JP6848290B2 - Image forming device and program - Google Patents

Description

The present invention relates to an image forming apparatus such as an MFP (Multi-Functional Peripheral) and related techniques.

There is a technique for discriminating between an area where a document exists (a document area) and an area where a document does not exist (an area outside the document) in a scanned image of a document placed on a platen, and erasing the area outside the document from the scanned image. To do.

For example, in the technique described in Patent Document 1, a frequency distribution of pixel values of a plurality of pixels in the entire scanned image is generated, and a threshold value for extracting edge candidates in a document region is calculated in the frequency distribution. Then, the original area and the outer area of the original are discriminated based on the threshold value calculated for the entire scanned image. When the document area and the document area are determined, the document area is deleted from the scanned image.

Japanese Unexamined Patent Publication No. 2003-60878

However, when a discrimination method (a method for discriminating between the original region and the outer region of the original) such as the technique described in Patent Document 1 is used, the original region and the outer region of the original are always separated in the generated scanned image. It is not always accurately determined.

For example, depending on the open / closed state of the document cover, when a discrimination method such as the technique described in Patent Document 1 is used, it may not be possible to accurately distinguish between the document area and the area outside the document in the generated scanned image. is there.

Both a light source that emits light and a light receiving unit that detects light are arranged below the glass surface for placing a document on the platen. Then, when the scanning process for the document placed on the platen is executed, light is emitted from the light source, and the light emitted from the light source and transmitted through the glass surface for placing the document hits the document on the platen and is reflected. The generated light passes through the glass surface for placing the document again and is detected by the light receiving unit.

Here, when the document cover is fully opened (for example, the document cover is opened to the extent that it is orthogonal to the platen) when the scanning process is executed, the document cover is emitted from the light source to cover the glass surface for placing the document. Of the transmitted light, the light that does not hit the document on the platen does not enter the light receiving unit and is not detected by the light receiving unit.

In this case, in the scanned image of the document (see FIG. 5), the pixel values of the pixels in the document area are relatively large (bright), and the pixel values of the pixels in the area outside the document are relatively small (dark). Then, with respect to such a scanned image, in the frequency distribution (frequency distribution relating to the pixel value of each pixel of the entire scanned image) generated by using the technique described in Patent Document 1, as shown in FIG. 6, comparison is made. Two peaks are formed, a peak having a targetly large brightness value and a peak having a relatively small brightness value. In the frequency distribution, the peak on the left side is the peak corresponding to the outside document region, and the peak on the right side is the peak corresponding to the document region. Then, for example, a threshold value for extracting edge candidates of the document region is set at a specific position (pixel value) of the base portion of the curve forming the peak corresponding to the region outside the document (peak on the left side).

As described above, when the document cover is fully opened, two peaks are formed in the frequency distribution for the entire scanned image, and an accurate threshold value is calculated. As a result, in the scanned image, the original area and the outer area of the original are accurately discriminated.

However, when the document cover is half-opened (about half-closed with respect to the fully-opened state) when the scanning process is executed, it may not be possible to accurately calculate the threshold value for the entire scanned image.

Specifically, when the document cover is in the half-open state as shown in FIG. 11, the light emitted from the light source and transmitted through the glass surface for placing the document, which does not hit the document on the document table, is the light of the document cover. When it hits the reflector, a part of the reflected light passes through the glass surface for placing the document again and is incident on the light receiving portion. In this case, a relatively large amount of reflected light is incident on one side of the platen (for example, the back side in the depth direction (the side close to the hinge of the document cover)), and almost no reflected light is incident on the other side (front side). .. Therefore, in the scanned image of the original, for example, as shown in FIG. 12, the pixel value on the back side (the portion where the incident of a relatively large amount of reflected light is detected) is relatively large (bright), and the position in the main scanning direction. However, a gradation is generated in which the pixel value gradually decreases (darkens) as the image moves from the back side to the front side.

When the scanned image has such a gradation, in the frequency distribution for the entire scanned image, it is caused by the fact that the pixel value of the pixel in the portion where a relatively large amount of reflected light is incident is relatively large among the pixels in the region outside the document. As shown in FIG. 13, two peaks, a peak corresponding to the document region and a peak corresponding to the outside region, are not formed (only a single peak is formed). In the frequency distribution having the single peak, it is difficult to specify the position (pixel value) to be set as the threshold value for extracting the edge candidate of the document region. Further, if the threshold value is set at a provisional position, the original area and the outer area of the original cannot be accurately discriminated in the scanned image. For example, when a threshold is tentatively set at the left foot of a single peak (see FIG. 13), both pixels in the manuscript area and the out-of-manuscript area are mixed in the single peak. In the scanned image, as shown in FIG. 23, the original area and the outer area of the original are not accurately discriminated.

As described above, depending on the open / closed state of the document cover, it may not be possible to accurately distinguish between the document area and the area outside the document in the scanned image when the discrimination method such as the technique described in Patent Document 1 is used. ..

Further, even in other situations, when a discrimination method such as the technique described in Patent Document 1 is used, it may not be possible to accurately discriminate between the original region and the non-manuscript region in the scanned image. That is, when a specific discrimination method is always used as a method for discriminating between the original region and the outer region of the original, it may not be possible to accurately discriminate between the original region and the outer region of the original in the scanned image.

Therefore, an object of the present invention is to provide a technique capable of more accurately discriminating between a document area and an area outside the document in a scanned image of a document placed on a platen.

In order to solve the above problems, the invention of claim 1 is an image processing apparatus, which comprises an analysis means for analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen. Based on the analysis result by the analysis means, from among a plurality of discrimination methods for discriminating between the manuscript area, which is the region where the manuscript exists in the image, and the extra-manuscript region, which is the region where the manuscript does not exist. The plurality of discrimination methods are provided with an area discriminating means for determining an optimum method which is the most suitable method for the scanned image and discriminating between the original area and the non-manuscript area by using the determined optimum method. Includes a first discrimination method and a third discrimination method, and the first discrimination method is a method of determining the peak position of a plurality of peaks existing in a frequency distribution relating to pixel values of a plurality of pixels in the entire scanned image. The peak having the smallest pixel value is determined as the peak outside the original region, which is the peak corresponding to the outer region of the original, and the remaining peak is determined as the peak in the original region, which is the peak corresponding to the outer region of the original. A value between the pixel value of the outer region peak and the pixel value of the original document region peak is set as a threshold value, and the original document region and the outer document region are discriminated based on the threshold value. , The scanned image is divided into one of the vertical direction and the horizontal direction and divided into a plurality of divided areas, and the frequency distribution for each divided area is the pixel value of a plurality of pixels in the divided area. Of the plurality of peaks existing in the frequency distribution with respect to, the value between the pixel value of the out-of-manuscript region peak and the pixel value of the manuscript region peak is set as the threshold value for each division region, and the threshold value for each division region is used as the threshold value. This is a method of discriminating between the original area and the outer area of the original for each divided area, and the analysis means determines the entire scanned image to a predetermined degree based on the spatial distribution of the pixel values of a plurality of pixels in the scanned image. It is determined whether or not the above gradation exists, and the area determination means determines the third determination method as the optimum method on condition that the gradation of the predetermined degree or more exists in the entire scanned image. The first discriminating method is determined as the optimum method on condition that there is no gradation of the predetermined degree or more in the entire scanned image.

The invention of claim 2 is the image processing apparatus according to the invention of claim 1, wherein when the scanned image has a gradation of the predetermined degree or more, the analysis means has the predetermined degree based on the spatial distribution situation. Further determining the change direction of the above gradation, the area determination means has a gradation of a predetermined degree or more in the entire scanned image and the change direction is either the vertical direction or the horizontal direction in the scanned image. On condition that it is determined that there is, the third discrimination method is determined as the optimum method, and the change direction of the gradation of a predetermined degree or more is determined as the one direction in the third discrimination method. It is characterized by.

The invention of claim 3 is the image processing apparatus according to the invention of claim 2, wherein the plurality of discrimination methods also include a fourth discrimination method, and the fourth discrimination method is provisional with respect to the scanned image. Edge candidates in the manuscript area are extracted by performing binarization processing and edge extraction processing based on a threshold value, and the manuscript area and the outside manuscript area are discriminated based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected, the four points are based on the positional relationship and the geometrical relationship of the detection positions of the remaining orthogonal points. The positions of at least one orthogonal point are estimated so that the closed region formed by connecting the orthogonal points of the above points with a straight line has a rectangular shape, the closed region is determined as the document region, and other than the closed region. The edge candidate of the document area is extracted by performing the method of discriminating the area as the outside area of the document and the binarization process and the edge extraction process based on a provisional threshold on the scanned image, and the extracted edge. Among the candidates, a straight line portion having a length equal to or longer than a predetermined value is specified as a regular edge portion of the document region, and a portion other than the straight line portion is specified as a false detection portion, and complement processing based on the regular edge portion is performed. Is performed to correct the erroneous detection portion, the closed region formed by the corrected edge candidate is determined as the document region, and the region other than the closed region is determined as the outside region. The fourth determination means, provided that the region determination means has a gradation of a predetermined degree or more in the entire scanned image and it is determined that the change direction is an oblique direction in the scanned image. The method is determined as the optimum method.

The invention of claim 4 is an image processing apparatus, which comprises an analysis means for analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen, and an analysis result by the analysis means. Based on this, the most suitable for the scanned image is among a plurality of discrimination methods for discriminating between a document area in which the document exists and an area outside the document in which the document does not exist in the image. The optimum method, which is a method, is determined, and the determined optimum method is used to provide an area discriminating means for discriminating between the manuscript area and the non-manuscript area. Including the third discrimination method, the first discrimination method selects the peak having the smallest pixel value at the peak position among a plurality of peaks existing in the frequency distribution regarding the pixel values of the plurality of pixels in the entire scanned image. , Estimate as the peak of the extra-manuscript region, which is the peak corresponding to the extra-manuscript region, and estimate the residual peak as the peak of the manuscript region, which is the peak corresponding to the extra-manuscript region. A method of discriminating between the manuscript area and the extra-manuscript region based on a value between the pixel values of the document region peak as a threshold, and the third discriminating method is a method of discriminating the scanned image in the vertical direction. It is divided into a plurality of divided regions by dividing it into one of the horizontal direction and the horizontal direction, and is a frequency distribution for each divided region and exists in the frequency distribution relating to the pixel values of a plurality of pixels in the divided region. The value between the pixel value of the out-of-manuscript region peak and the pixel value of the out-of-manuscript region peak is set as the threshold value for each division region, and the manuscript region and the out-of-manuscript region are based on the threshold value for each division region. Is a method of discriminating between the two for each divided region, and the analysis means has a constant distance between the out-of-manuscript region peak and the manuscript region peak in the frequency distribution relating to the pixel values of a plurality of pixels in the entire scanned image. The region determination means determines that the distance between the extra-manuscript region peak and the manuscript region peak is larger than a certain degree in the frequency distribution, provided that the region determination means is larger than the above-mentioned constant. The first discrimination method is determined as the optimum method, and the analysis means has a gradation of a predetermined degree or more in the entire scanned image based on the spatial distribution of pixel values of a plurality of pixels in the scanned image. The area determination means has a gradation of a predetermined degree or more in the entire scanned image. Is a condition, and the third discrimination method is determined as the optimum method.

The invention of claim 5 is the image processing apparatus according to the invention of claim 4, wherein the plurality of discrimination methods also include a second discrimination method, and the second discrimination method is the amount of light received by the light receiving element in the scanning process. This is an input / output characteristic adjustment process for adjusting the input / output characteristics between the analog input value indicating the output pixel value and the digital output value indicating the output pixel value, and is a part of the output range in the input / output characteristics before the input / output characteristic adjustment process. Adjusting the input / output characteristics so that the range becomes a wider output range Newly required in the frequency distribution regarding the pixel values of a plurality of pixels in the entire second scan image obtained after the input / output characteristic adjustment processing is performed. A method of discriminating between the manuscript area and the out-of-manuscript region based on a value between the pixel value of the out-of-manuscript region peak and the newly obtained pixel value of the manuscript region peak as a threshold, and the above The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates of the document area, and the document area is based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected, the positional relationship and the geometrical relationship of the detection positions of the remaining orthogonal points are determined. Based on the above, the positions of the at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is used as the manuscript region. The edge candidate of the document region is subjected to a method of discriminating the region other than the closed region as the outside region of the document, and binarization processing and edge extraction processing based on a provisional threshold value on the scanned image. Is extracted, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value is specified as a regular edge portion of the original document region, and a portion other than the straight line portion is specified as a false detection portion. Complementary processing based on the regular edge portion is performed to correct the false detection portion, the closed region formed by the corrected edge candidate is determined as the document region, and the region other than the closed region is the outside region of the document. The second determination means on the condition that it is determined that the distance between the out-of-manuscript region peak and the manuscript region peak is smaller than a certain degree. The method is determined as the optimum method.

The invention of claim 6 is the image processing apparatus according to the invention of claim 4, wherein when the scanned image has a gradation of the predetermined degree or more, the analysis means has the predetermined degree based on the spatial distribution situation. Further determining the change direction of the above gradation, the area determination means has a gradation of a predetermined degree or more in the entire scanned image and the change direction is either the vertical direction or the horizontal direction in the scanned image. On condition that it is determined that there is, the third discrimination method is determined as the optimum method, and the change direction of the gradation of a predetermined degree or more is determined as the one direction in the third discrimination method. It is characterized by.

The invention of claim 7 is the image processing apparatus according to the invention of claim 6, wherein the plurality of discrimination methods also include a fourth discrimination method, and the fourth discrimination method is provisional with respect to the scanned image. Edge candidates in the manuscript area are extracted by performing binarization processing and edge extraction processing based on a threshold value, and the manuscript area and the outside manuscript area are discriminated based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected, the four points are based on the positional relationship and the geometrical relationship of the detection positions of the remaining orthogonal points. The positions of at least one orthogonal point are estimated so that the closed region formed by connecting the orthogonal points of the above points with a straight line has a rectangular shape, the closed region is determined as the document region, and other than the closed region. The edge candidate of the document area is extracted by performing the method of discriminating the area as the outside area of the document and the binarization process and the edge extraction process based on a provisional threshold on the scanned image, and the extracted edge. Among the candidates, a straight line portion having a length equal to or longer than a predetermined value is specified as a regular edge portion of the document region, and a portion other than the straight line portion is specified as a false detection portion, and complementary processing based on the regular edge portion is performed. Is performed to correct the erroneous detection portion, the closed region formed by the corrected edge candidate is determined as the document region, and the region other than the closed region is determined as the outside region. The fourth determination means, provided that the region determination means has a gradation of a predetermined degree or more in the entire scanned image and it is determined that the change direction is an oblique direction in the scanned image. The method is determined as the optimum method.

The invention of claim 8 is the image processing apparatus according to the invention of claim 7, wherein the plurality of discrimination methods also include a second discrimination method, and the second discrimination method is the amount of light received by the light receiving element in the scanning process. This is an input / output characteristic adjustment process for adjusting the input / output characteristics between the analog input value indicating the output pixel value and the digital output value indicating the output pixel value, and is a part of the output range in the input / output characteristics before the input / output characteristic adjustment process. Adjusting the input / output characteristics so that the range becomes a wider output range Newly required in the frequency distribution regarding the pixel values of a plurality of pixels in the entire second scan image obtained after the input / output characteristic adjustment processing is performed. A method of discriminating between the manuscript area and the out-of-manuscript region based on a value between the pixel value of the out-of-manuscript region peak and the newly obtained pixel value of the out-of-manuscript region peak as a threshold, and the above. The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates of the document area, and the document area is based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected, the positional relationship and the geometrical relationship of the detection positions of the remaining orthogonal points are determined. Based on the above, the positions of the at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is used as the manuscript region. The edge candidate of the document region is subjected to a method of discriminating the region other than the closed region as the outside region of the document, and binarization processing and edge extraction processing based on a provisional threshold value on the scanned image. Is extracted, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value is specified as a regular edge portion of the original document region, and a portion other than the straight line portion is specified as a false detection portion. Complementary processing based on the regular edge portion is performed to correct the false detection portion, the closed region formed by the corrected edge candidate is determined as the document region, and the region other than the closed region is the outside region of the document. The second determination means on the condition that it is determined that the distance between the out-of-manuscript region peak and the manuscript region peak is smaller than a certain degree. The method is determined as the optimum method.

The invention of claim 9 is characterized in that, in the image processing apparatus according to the invention of claim 8, the second discrimination method is the same as the fourth discrimination method.

The invention of claim 10 is characterized in that, in the image processing apparatus according to the invention of claim 8, the second discrimination method is different from the fourth discrimination method.

The invention of claim 11 is an image processing apparatus, which comprises an analysis means for analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen, and an analysis result by the analysis means. Based on this, the most suitable for the scanned image is among a plurality of discrimination methods for discriminating between a document area in which the document exists and an area outside the document in which the document does not exist in the image. The optimum method, which is a method, is determined, and the determined optimum method is used to provide an area discriminating means for discriminating between the manuscript area and the non-manuscript area. Including the fourth discrimination method, the third discrimination method divides the scanned image into one of the vertical direction and the horizontal direction, divides the scanned image into a plurality of divided regions, and distributes the frequency for each divided region. Of the plurality of peaks existing in the frequency distribution relating to the pixel values of the plurality of pixels in the divided region, the pixel value of the extra-manuscript region peak, which is the peak corresponding to the extra-manuscript region, and the peak corresponding to the manuscript region. This is a method in which a value between the pixel value of the original document area peak and the pixel value of the original document area peak is set as a threshold value for each divided area, and the original document area and the outer document area are discriminated for each divided area based on the threshold value for each divided area. In the fourth discrimination method, edge candidates in the manuscript region are extracted by performing binarization processing and edge extraction processing based on a provisional threshold on the scanned image, and the edge candidates are orthogonal to each other at four points. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the original region from the outer region based on the position of the points, the remaining orthogonal points are detected. Based on the positional relationship and the geometric relationship of the positions, the positions of the at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape. , A method of discriminating the closed area as the original area and determining an area other than the closed area as the outer area of the original, and binarization processing and edge extraction processing based on a provisional threshold value for the scanned image. The edge candidates of the manuscript area are extracted, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value is specified as a regular edge portion of the manuscript region, and a portion other than the straight line portion is specified. It is specified as a false detection portion, complementary processing based on the regular edge portion is performed to correct the false detection portion, a closed region formed by the corrected edge candidate is determined as the original document region, and the closed region is determined. Other than It is one of the methods of discriminating the region as the region outside the original document, and the analysis means has a predetermined degree or more over the entire scanned image based on the spatial distribution of the pixel values of a plurality of pixels in the scanned image. It is determined whether or not there is a gradation, and if the scanned image has a gradation of the predetermined degree or more, the change direction of the gradation of the predetermined degree or more is further determined based on the spatial distribution situation, and the region. The discriminating means is the first on the condition that it is determined that there is a gradation of a predetermined degree or more in the entire scanned image and the change direction is either the vertical direction or the horizontal direction in the scanned image. The three discrimination methods are determined as the optimum method, and the change direction of the gradation of the predetermined degree or more is determined as the one direction in the third discrimination method, and the gradation of the predetermined degree or more exists in the entire scanned image. Moreover, the fourth determination method is determined as the optimum method on condition that it is determined that the change direction is the oblique direction in the scanned image.

The invention of claim 12 is the image processing apparatus according to any one of claims 1 to 11, wherein the third discrimination method uses the scanned image in one of the vertical direction and the horizontal direction. In addition to dividing into the plurality of divided regions, peaks having a frequency distribution of 2 or more with respect to pixel values of a plurality of pixels in each of the divided regions in two or more divided regions are defined as Of the plurality of peaks existing in each frequency distribution with respect to the two or more divided regions having the common position peak, which is a peak having the peak position at the common position in the two or more frequency distributions related to the two or more divided regions. It is determined as the original region peak, and a specific non-common position peak among the non-common position peaks other than the common position peak is determined as the extra-manuscript region peak, and the pixel value of the extra-manuscript region peak and the said A feature of this method is that a value between the pixel value of the peak of the document area is set as a threshold value for each divided area, and the document area and the outside area of the document are discriminated for each divided area based on the threshold value for each divided area. And.

According to a thirteenth aspect of the present invention, in the image processing apparatus according to the twelfth aspect of the present invention, in the third discrimination method, the frequency distribution relating to the pixel values of the plurality of pixels of the plurality of divided regions has a peak of 2 or more. The threshold value of the divided region that does not have is obtained based on the threshold value of the other divided region.

The invention of claim 14 is an image processing apparatus, which comprises an analysis means for analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen, and an analysis result by the analysis means. Based on this, the most suitable for the scanned image is among a plurality of discrimination methods for discriminating between a document area in which the document exists and an area outside the document in which the document does not exist in the image. The optimum method, which is a method, is determined, and the determined optimum method is used to provide an area discriminating means for discriminating between the manuscript area and the non-manuscript area. Including the second discrimination method, the first discrimination method selects the peak having the smallest pixel value at the peak position among a plurality of peaks existing in the frequency distribution regarding the pixel values of the plurality of pixels in the entire scanned image. , It is estimated as the peak of the extra-manuscript region, which is the peak corresponding to the extra-manuscript region, and the residual peak is estimated as the peak of the manuscript region, which is the peak corresponding to the extra-manuscript region. A method of discriminating between the manuscript region and the non-manuscript region based on a value between the pixel values of the document region peak as a threshold, and the second discriminating method is a light receiving element in the scanning process. It is an input / output characteristic adjustment process that adjusts the input / output characteristics between the analog input value indicating the amount of light received and the digital output value indicating the output pixel value, and is the output range in the input / output characteristics before the input / output characteristic adjustment process. Adjusting the input / output characteristics so that a part of the range becomes a wider output range New in the frequency distribution regarding the pixel values of multiple pixels in the entire second scan image obtained after the input / output characteristic adjustment processing is performed. A method of discriminating between the manuscript area and the out-of-manuscript region based on a value between the pixel value of the out-of-manuscript region peak obtained in 1 and the newly obtained pixel value of the out-of-manuscript region peak as a threshold value. Then, the scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates in the original area, and based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the original area from the outer area of the original, the positional relationship between the detection positions of the remaining orthogonal points and the residual orthogonal points. Based on the geometrical relationship, the at least one orthogonal point has a rectangular shape so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape. A method of estimating the position, determining the closed area as the original area, and determining an area other than the closed area as the outer area of the original, and binarization processing based on a provisional threshold value for the scanned image. Edge extraction processing is performed to extract edge candidates of the manuscript area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value is specified as a regular edge portion of the manuscript region, and the straight line portion is specified. A portion other than the false detection portion is specified as a false detection portion, the false detection portion is corrected by performing complementary processing based on the normal edge portion, and a closed region formed by the corrected edge candidate is determined as the original document region. It is one of the methods of discriminating the region other than the closed region as the extra-manuscript region, and the region discriminating means is the extra-manuscript region in the frequency distribution relating to the pixel values of a plurality of pixels in the entire scanned image. The first discrimination method is determined as the optimum method on condition that it is determined that the distance between the peak and the original document area peak is larger than a certain degree, and the distance between the outer document area peak and the original document area peak is determined. Is determined as the optimum method on the condition that it is determined that is smaller than a certain degree.

The invention of claim 15 is an image processing apparatus, which comprises an analysis means for analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen, and an analysis result by the analysis means. Based on this, the most suitable for the scanned image is among a plurality of discrimination methods for discriminating between a document area in which the document exists and an area outside the document in which the document does not exist in the image. The optimum method, which is a method, is determined, and the determined optimum method is used to provide an area discriminating means for discriminating between the manuscript area and the non-manuscript area. Including the fourth discrimination method, the first discrimination method selects the peak having the smallest pixel value at the peak position among a plurality of peaks existing in the frequency distribution relating to the pixel values of the plurality of pixels in the entire scanned image. , It is determined as the peak of the extra-manuscript region, which is the peak corresponding to the extra-manuscript region, and the residual peak is determined as the peak of the original region, which is the peak corresponding to the extra-manuscript region. A method of discriminating between the manuscript region and the extra-manuscript region based on a value between the pixel values of the document region peak as a threshold, and the fourth discriminating method is for the scanned image. Edge candidates in the manuscript area are extracted by performing binarization processing and edge extraction processing based on a provisional threshold, and the manuscript area and the outer manuscript area are based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected, based on the positional relationship and the geometric relationship of the detection positions of the remaining orthogonal points, The positions of the at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is determined as the manuscript region. Edge candidates of the original area are extracted and extracted by performing a method of discriminating an area other than the closed area as the outer area of the original, and performing binarization processing and edge extraction processing based on a provisional threshold on the scanned image. Of the edge candidates, a straight line portion having a length equal to or longer than a predetermined value is specified as a regular edge portion of the document region, and a portion other than the straight line portion is specified as a false detection portion, and the regular edge portion is used. Complementing processing based on the above is performed to correct the false detection portion, a closed region formed by the corrected edge candidate is determined as the original document region, and an region other than the closed region is the original document. It is one of the methods of discriminating as an outer region, and the analysis means has a gradation of a predetermined degree or more in the entire scanned image based on the spatial distribution of the pixel values of a plurality of pixels in the scanned image. The area determination means determines whether or not the fourth determination method is the optimum method on condition that a gradation of a predetermined degree or more is present in the entire scan image, and the region determination means is used in the entire scan image. It is characterized in that the first determination method is determined as the optimum method on condition that a gradation of a predetermined degree or more does not exist.

The invention of claim 16 is an image processing apparatus, which comprises an analysis means for analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen, and an analysis result by the analysis means. Based on this, the most suitable for the scanned image is among a plurality of discrimination methods for discriminating between a document area in which the document exists and an area outside the document in which the document does not exist in the image. The optimum method, which is a method, is determined, and the determined optimum method is used to provide an area discriminating means for discriminating between the manuscript area and the non-manuscript area. Including the fourth discrimination method, the first discrimination method selects the peak having the smallest pixel value at the peak position among a plurality of peaks existing in the frequency distribution relating to the pixel values of the plurality of pixels in the entire scanned image. , It is estimated as the peak of the extra-manuscript region, which is the peak corresponding to the extra-manuscript region, and the residual peak is estimated as the peak of the manuscript region, which is the peak corresponding to the extra-manuscript region. A method of discriminating between the manuscript region and the extra-manuscript region based on a value between the pixel values of the document region peak as a threshold, and the fourth discriminating method is for the scanned image. Edge candidates in the manuscript area are extracted by performing binarization processing and edge extraction processing based on a provisional threshold, and the manuscript area and the outer manuscript area are based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected, based on the positional relationship and the geometric relationship of the detection positions of the remaining orthogonal points, The positions of the at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is determined as the manuscript region. Edge candidates of the original area are extracted and extracted by performing a method of discriminating an area other than the closed area as the outer area of the original, and performing binarization processing and edge extraction processing based on a provisional threshold on the scanned image. Of the edge candidates, a straight line portion having a length equal to or longer than a predetermined value is specified as a regular edge portion of the document region, and a portion other than the straight line portion is specified as a false detection portion, and the regular edge portion is used. Complementing processing based on the above is performed to correct the false detection portion, a closed region formed by the corrected edge candidate is determined as the original document region, and an region other than the closed region is the original document. It is one of the methods of discriminating as an outer region, and the analysis means has a constant distance between the outer region peak of the original document and the peak of the original document in the frequency distribution regarding the pixel values of a plurality of pixels in the entire scanned image. It is a condition that the area determination means determines in the frequency distribution that the distance between the extra-manuscript region peak and the manuscript region peak is larger than the certain degree. The first discrimination method is determined as the optimum method, and the analysis means has a gradation of a predetermined degree or more over the entire scanned image based on the spatial distribution of pixel values of a plurality of pixels in the scanned image. The area determination means determines whether or not it exists, and determines the fourth determination method as the optimum method on condition that a gradation of a predetermined degree or more exists in the entire scanned image. To do.

The invention of claim 17 is an image processing apparatus, which comprises an analysis means for analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen, and an analysis result by the analysis means. Based on this, the most suitable for the scanned image is among a plurality of discrimination methods for discriminating between a document area in which the document exists and an area outside the document in which the document does not exist in the image. The optimum method, which is a method, is determined, and the determined optimum method is used to provide an area discriminating means for discriminating between the manuscript area and the non-manuscript area. Including the fourth discrimination method, the second discrimination method adjusts the input / output characteristics between the analog input value indicating the amount of light received by the light receiving element in the scanning process and the digital output value indicating the output pixel value. After the input / output characteristic adjustment processing that adjusts the input / output characteristics so that a part of the output range in the input / output characteristics before the input / output characteristic adjustment processing becomes a wider output range in the characteristic adjustment processing In the frequency distribution relating to the pixel values of a plurality of pixels in the entire obtained second scan image, the peak having the smallest pixel value at the peak position among the plurality of peaks is the peak corresponding to the out-of-manuscript region. It is determined as a region peak, the residual peak is determined as the original region peak which is the peak corresponding to the original region, and the value between the pixel value of the extra-manuscript region peak and the pixel value of the original region peak is determined. It is a method of discriminating between the manuscript area and the extra-manuscript area based on the threshold value, and the fourth discriminating method is a binarization process and edge extraction based on a provisional threshold value for the scanned image. In performing the process to extract the edge candidates of the manuscript region and discriminating between the manuscript region and the outer region of the manuscript based on the positions of the four orthogonal points where the edge candidates are orthogonal to each other, the four orthogonal points When it is determined that at least one of the orthogonal points is not detected, the four orthogonal points are connected by a straight line based on the positional relationship and the geometric relationship of the detection positions of the remaining orthogonal points. The positions of the at least one orthogonal point are estimated so that the closed region to be formed has a rectangular shape, the closed region is determined as the document region, and the region other than the closed region is determined as the outside region. The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold value to extract edge candidates in the original area, and the extracted edge candidates are extracted. Then, a straight line portion having a length equal to or longer than a predetermined value is specified as a regular edge portion of the document region, and a portion other than the straight line portion is specified as a false detection portion, and complement processing is performed based on the regular edge portion. It is one of the methods of correcting the erroneous detection portion, determining the closed region formed by the corrected edge candidate as the document region, and determining the region other than the closed region as the outer region of the document. The analysis means determines whether or not there is a gradation of a predetermined degree or more in the entire scan image based on the spatial distribution of the pixel values of the plurality of pixels in the scan image, and the area determination means The fourth discrimination method is determined as the optimum method on the condition that the gradation of the predetermined degree or more exists in the entire scanned image, and the condition that the gradation of the predetermined degree or more does not exist in the entire scanned image. The second discrimination method is determined as the optimum method.

The invention of claim 18 is based on a) a step of analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen, and b) an analysis result in the step a). Based on this, the most suitable for the scanned image is among a plurality of discrimination methods for discriminating between a document area in which the document exists and an area outside the document in which the document does not exist in the image. A program for executing a step of determining the optimum method, which is a method, and a step of c) discriminating between the manuscript area and the non-manuscript area using the optimum method determined in step b). The plurality of discrimination methods include a first discrimination method and a third discrimination method, and the first discrimination method includes a plurality of discrimination methods existing in a frequency distribution relating to pixel values of a plurality of pixels in the entire scanned image. Of the peaks, the peak with the smallest pixel value at the peak position is estimated as the peak outside the document, which is the peak corresponding to the outside region, and the remaining peak is the peak corresponding to the outside region, which is the original region. In addition to estimating as a peak, a value between the pixel value of the extra-manuscript region peak and the pixel value of the manuscript region peak is used as a threshold, and the manuscript region and the extra-manuscript region are discriminated based on the threshold. In the third discrimination method, the scanned image is divided into one of the vertical direction and the horizontal direction and divided into a plurality of divided regions, and the frequency distribution for each divided region is the divided region. Of the plurality of peaks existing in the frequency distribution relating to the pixel values of the plurality of pixels, the value between the pixel value of the out-of-manuscript region peak and the pixel value of the manuscript region peak is set as the threshold value for each division region, and the division region. It is a method of discriminating the original area and the outer area of the original for each divided area based on the threshold value for each, and the step a) is a-1) the space of the pixel values of a plurality of pixels in the scanned image. A step of determining whether or not a predetermined degree or more of gradation exists in the entire scanned image based on a distribution situation, and the step b) is b-1) the predetermined degree or more in the entire scanned image. The step of determining the third determination method as the optimum method on condition that the above-mentioned gradation is present, and b-2) the first, on condition that there is no gradation of the predetermined degree or more in the entire scanned image. 1 It is characterized by having a step of determining a discrimination method as the optimum method.
According to the invention of claim 19, in the program according to the invention of claim 18, the step a) is a-2) when the scanned image has a gradation of a predetermined degree or more, based on the spatial distribution situation. Further comprising a step of further determining the change direction of the gradation of the predetermined degree or more, and in the step b-1), the gradation of the predetermined degree or more exists in the entire scanned image and the change direction is the scan image. The third discrimination method is determined as the optimum method on condition that it is determined to be either the vertical direction or the horizontal direction in the above, and the change direction of the gradation of a predetermined degree or more is the first. 3 It is characterized in that it is determined as the one direction in the discrimination method.
The invention of claim 20 is the program according to the invention of claim 19, wherein the plurality of discrimination methods also include a fourth discrimination method, and the fourth discrimination method sets a provisional threshold value for the scanned image. In extracting edge candidates of the manuscript region by performing binarization processing and edge extraction processing based on the above, and discriminating between the manuscript region and the outer region of the manuscript based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected, the four points are orthogonal based on the positional relationship and the geometrical relationship of the detection positions of the remaining orthogonal points. The positions of the at least one orthogonal point are estimated so that the closed region formed by connecting the points with a straight line has a rectangular shape, the closed region is determined as the document region, and the region other than the closed region is determined. Is extracted as an edge candidate of the document area by performing a method of discriminating as the outside area of the document and a binarization process and an edge extraction process based on a provisional threshold on the scanned image, and the extracted edge candidate Of these, a straight line portion having a length equal to or longer than a predetermined value is specified as a regular edge portion of the document region, and a portion other than the straight line portion is specified as a false detection portion, and complement processing is performed based on the regular edge portion. It is one of the methods of correcting the erroneous detection portion, determining the closed region formed by the corrected edge candidate as the document region, and determining the region other than the closed region as the outside region. The step b) is the first on the condition that it is determined that there is a gradation of a predetermined degree or more in the entire scanned image and the change direction is an oblique direction in the scanned image. 4 It is characterized by further having a step of determining the discrimination method as the optimum method.
The invention of claim 21 is based on a) a step of analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen, and b) an analysis result in the step a). Based on this, the most suitable for the scanned image is among a plurality of discrimination methods for discriminating between a document area in which the document exists and an area outside the document in which the document does not exist in the image. A program for executing a step of determining the optimum method, which is a method, and a step of c) discriminating between the manuscript area and the non-manuscript area using the optimum method determined in step b). The plurality of discrimination methods include a first discrimination method and a third discrimination method, and the first discrimination method includes a plurality of discrimination methods existing in a frequency distribution relating to pixel values of a plurality of pixels in the entire scanned image. Among the peaks, the peak having the smallest pixel value at the peak position is estimated as the extra-manuscript region peak which is the peak corresponding to the extra-manuscript region, and the remaining peak is the manuscript region which is the peak corresponding to the extra-manuscript region. In addition to estimating as a peak, a value between the pixel value of the extra-manuscript region peak and the pixel value of the manuscript region peak is used as a threshold, and the manuscript region and the extra-manuscript region are discriminated based on the threshold. In the third discrimination method, the scanned image is divided into one of the vertical direction and the horizontal direction and divided into a plurality of divided regions, and the frequency distribution for each divided region is the divided region. Of the plurality of peaks existing in the frequency distribution relating to the pixel values of the plurality of pixels, the value between the pixel value of the out-of-manuscript region peak and the pixel value of the manuscript region peak is set as the threshold value for each division region, and the division region. It is a method of discriminating the original area and the outer area of the original for each divided area based on the threshold value for each, and the step a) relates to a-1) pixel values of a plurality of pixels in the entire scanned image. In the frequency distribution, a step of determining whether or not the distance between the out-of-manuscript region peak and the manuscript region peak is larger than a certain degree, and a-2) spatial distribution of pixel values of a plurality of pixels in the scanned image. Based on the situation, the scan image has a step of determining whether or not there is a gradation of a predetermined degree or more in the entire scanned image, and the step b) is b-1) the extra-manuscript region in the frequency distribution. The first determination method is performed on condition that it is determined that the distance between the peak and the original region peak is larger than the certain degree. It has a step of determining as the optimum method, and b-2) a step of determining the third discrimination method as the optimum method on condition that a gradation of a predetermined degree or more exists in the entire scanned image. It is a feature.
The invention of claim 22 is the program according to the invention of claim 21, wherein the plurality of discrimination methods also include a second discrimination method, and the second discrimination method indicates the amount of light received by the light receiving element in the scanning process. It is an input / output characteristic adjustment process that adjusts the input / output characteristics between the analog input value and the digital output value indicating the output pixel value, and a part of the output range in the input / output characteristics before the input / output characteristic adjustment process is The manuscript newly obtained in the frequency distribution relating to the pixel values of a plurality of pixels in the entire second scan image obtained after the input / output characteristic adjustment processing for adjusting the input / output characteristics so as to have a wider output range is performed. A method of discriminating between the original region and the outer region of the original based on a value between the pixel value of the peak of the outer region and the newly obtained pixel value of the peak of the original region as a threshold, and the scanned image. The edge candidates of the manuscript area are extracted by performing binarization processing and edge extraction processing based on a provisional threshold value, and the manuscript area and the above are based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the area from the outside of the document, the positional relationship and the geometric relationship of the detection positions of the remaining orthogonal points are determined. Based on this, the positions of the at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is determined as the manuscript region. Then, an edge candidate of the document area is extracted by performing a method of discriminating the area other than the closed area as the outside area of the document, and performing binarization processing and edge extraction processing based on a provisional threshold value on the scanned image. Then, among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value is specified as a regular edge portion of the original document region, and a portion other than the straight line portion is specified as a false detection portion, and the normal portion is specified. Complementing processing based on the edge portion is performed to correct the false detection portion, the closed region formed by the corrected edge candidate is determined as the original region, and the region other than the closed region is determined as the outer region of the original. In step b), it is determined that the distance between the out-of-manuscript region peak and the manuscript region peak is smaller than a certain level. 2. It is characterized by further having a step of determining the discrimination method as the optimum method.
The invention of claim 23 is based on the spatial distribution situation in the program according to the invention of claim 21, when the step a) is a-3) the scan image has a gradation of a predetermined degree or more. Further comprising a step of further determining the change direction of the gradation of the predetermined degree or more, and in the step b-2), the gradation of the predetermined degree or more exists in the entire scanned image and the change direction is the scanned image. The third discrimination method is determined as the optimum method on condition that it is determined to be either the vertical direction or the horizontal direction in the above, and the change direction of the gradation of a predetermined degree or more is the first. 3 It is characterized in that it is determined as the one direction in the discrimination method.
The invention of claim 24 is the program according to the invention of claim 23, wherein the plurality of discrimination methods also include a fourth discrimination method, and the fourth discrimination method sets a provisional threshold value for the scanned image. In extracting edge candidates in the manuscript area by performing binarization processing and edge extraction processing based on the above, and discriminating between the manuscript area and the outside manuscript area based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected, the four points are orthogonal based on the positional relationship and the geometric relationship of the detection positions of the remaining orthogonal points. The positions of the at least one orthogonal point are estimated so that the closed region formed by connecting the points with a straight line has a rectangular shape, the closed region is determined as the document region, and the region other than the closed region is determined. Is extracted as an edge candidate of the document area by performing a method of discriminating as the outside area of the document and a binarization process and an edge extraction process based on a provisional threshold on the scanned image, and the extracted edge candidate Of these, a straight line portion having a length equal to or longer than a predetermined value is specified as a regular edge portion of the document region, and a portion other than the straight line portion is specified as a false detection portion, and complement processing is performed based on the regular edge portion. It is one of the methods of correcting the erroneous detection portion, determining the closed region formed by the corrected edge candidate as the original document region, and determining the region other than the closed region as the outer document region. In step b), the first step is b-4) on condition that it is determined that a gradation of a predetermined degree or more exists in the entire scanned image and the change direction is an oblique direction in the scanned image. 4 It is characterized by further having a step of determining the discrimination method as the optimum method.
The invention of claim 25 includes the second discrimination method in the program according to the invention of claim 24, and the second discrimination method indicates the amount of light received by the light receiving element in the scanning process. It is an input / output characteristic adjustment process that adjusts the input / output characteristics between the analog input value and the digital output value indicating the output pixel value, and a part of the output range in the input / output characteristics before the input / output characteristic adjustment process is The manuscript newly obtained in the frequency distribution relating to the pixel values of a plurality of pixels in the entire second scan image obtained after the input / output characteristic adjustment processing for adjusting the input / output characteristics so as to have a wider output range is performed. A method of discriminating between the original region and the outer region of the original based on a value between the pixel value of the peak of the outer region and the newly obtained pixel value of the peak of the original region as a threshold, and the scanned image. The edge candidates of the manuscript area are extracted by performing binarization processing and edge extraction processing based on a provisional threshold value, and the manuscript area and the above are based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the area from the outside of the document, the positional relationship and the geometric relationship of the detection positions of the remaining orthogonal points are determined. Based on this, the positions of the at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is determined as the manuscript region. Then, an edge candidate of the document area is extracted by performing a method of discriminating the area other than the closed area as the outside area of the document, and performing binarization processing and edge extraction processing based on a provisional threshold value on the scanned image. Then, among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value is specified as a regular edge portion of the original document region, and a portion other than the straight line portion is specified as a false detection portion, and the normal portion is specified. Complementing processing based on the edge portion is performed to correct the false detection portion, the closed region formed by the corrected edge candidate is determined as the original region, and the region other than the closed region is determined as the outer region of the original. In step b), it is determined that the distance between the out-of-manuscript region peak and the manuscript region peak is smaller than a certain level. 2. It is characterized by further having a step of determining the discrimination method as the optimum method.
The invention of claim 26 is characterized in that, in the program according to the invention of claim 25, the second discrimination method is the same method as the fourth discrimination method.
The invention of claim 27 is characterized in that, in the program according to the invention of claim 25, the second discrimination method is a method different from the fourth discrimination method.
The invention of claim 28 is based on a) a step of analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen, and b) an analysis result in the step a). Based on this, the most suitable for the scanned image is among a plurality of discrimination methods for discriminating between a document area in which the document exists and an area outside the document in which the document does not exist in the image. A program for executing a step of determining the optimum method, which is a method, and a step of c) discriminating between the manuscript area and the non-manuscript area using the optimum method determined in step b). The plurality of discrimination methods include a third discrimination method and a fourth discrimination method, and the third discrimination method divides the scanned image into one of a vertical direction and a horizontal direction. It is divided into a plurality of divided regions, and is a peak corresponding to the extra-manuscript region among a plurality of peaks existing in the frequency distribution related to the pixel values of a plurality of pixels in the divided region, which is a frequency distribution for each divided region. The value between the pixel value of the peak outside the document area and the pixel value of the peak document area, which is the peak corresponding to the document area, is set as the threshold value for each divided area, and the document area and the document area are based on the threshold value for each divided area. It is a method of discriminating the outer region of the original for each divided region, and the fourth discriminating method performs a binarization process and an edge extraction process based on a provisional threshold value on the scanned image to perform an edge extraction process of the document region. In extracting candidates and discriminating between the manuscript region and the outside-manuscript region based on the positions of the four orthogonal points where the edge candidates are orthogonal to each other, at least one of the four orthogonal points is the orthogonal point. When it is determined that the detection is not detected, the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape based on the positional relationship and the geometric relationship of the detection positions of the remaining orthogonal points. As described above, a method of estimating the positions of at least one orthogonal point, determining the closed region as the original region, and determining an region other than the closed region as the outer region of the original, and the scanned image. Edge candidates of the manuscript area are extracted by performing binarization processing and edge extraction processing based on a provisional threshold, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value is a normal part of the manuscript area. The part other than the straight line part is specified as an erroneous detection part, and the complement processing based on the regular edge part is performed to correct the erroneous detection part, and the corrected d. It is one of the methods of discriminating a closed region formed by a gradation candidate as the manuscript region and discriminating a region other than the closed region as the outer region of the manuscript. Based on the spatial distribution of the pixel values of a plurality of pixels in the scanned image, a step of determining whether or not there is a gradation of a predetermined degree or more in the entire scanned image, and a-2) the predetermined in the scanned image. When there is a gradation of a degree or more, the step b) includes a step of further determining the change direction of the gradation of the predetermined degree or more based on the spatial distribution situation, and the step b) is b-1) the scanned image. The third discrimination method is the optimum method on condition that it is determined that there is a gradation of a predetermined degree or more in the whole and that the change direction is either the vertical direction or the horizontal direction in the scanned image. There are a step of determining the change direction of the gradation of the predetermined degree or more as the one direction in the third discrimination method, and b-2) the gradation of the predetermined degree or more in the entire scanned image. Further, it is characterized by having a step of determining the fourth determination method as the optimum method on condition that it is determined that the change direction is an oblique direction in the scanned image.
The invention of claim 29 is a program according to any one of claims 18 to 28, wherein the third discrimination method classifies the scanned image into one of a vertical direction and a horizontal direction. In addition to being divided into a plurality of divided regions, two or more divided regions of the plurality of divided regions having a frequency distribution of two or more peaks with respect to pixel values of a plurality of pixels in each divided region. Regarding the divided region, among the plurality of peaks existing in each frequency distribution, the common position peak which is the peak having the peak position at the common position in the two or more frequency distributions related to the two or more divided regions is the document region peak. Of the non-common position peaks that are peaks other than the common position peak, a specific non-common position peak is determined as the extra-manuscript region peak, and the pixel value of the extra-manuscript region peak and the original document region peak. A value between the pixel value and the pixel value is set as a threshold value for each divided region, and the method is characterized in that the document region and the outer region of the document are discriminated for each divided region based on the threshold value for each divided region.
According to the invention of claim 30, in the program according to the invention of claim 29, in the third discrimination method, the frequency distribution relating to the pixel values of the plurality of pixels of the plurality of divided regions does not have two or more peaks. The threshold value of the divided region is obtained based on the threshold value of another divided region.
The invention of claim 31 is based on a) a step of analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen, and b) an analysis result in the step a). Based on this, the most suitable for the scanned image is among a plurality of discrimination methods for discriminating between a document area in which the document exists and an area outside the document in which the document does not exist in the image. A program for executing a step of determining the optimum method, which is a method, and a step of c) discriminating between the manuscript area and the non-manuscript area using the optimum method determined in step b). The plurality of discrimination methods include a first discrimination method and a second discrimination method, and the first discrimination method includes a plurality of discrimination methods existing in a frequency distribution relating to pixel values of a plurality of pixels in the entire scanned image. Among the peaks, the peak having the smallest pixel value at the peak position is estimated as the extra-manuscript region peak which is the peak corresponding to the extra-manuscript region, and the remaining peak is the manuscript region which is the peak corresponding to the extra-manuscript region. In addition to estimating as a peak, a value between the pixel value of the extra-manuscript region peak and the pixel value of the manuscript region peak is used as a threshold, and the manuscript region and the extra-manuscript region are discriminated based on the threshold. The second discrimination method is an input / output characteristic adjustment process for adjusting an input / output characteristic between an analog input value indicating the amount of light received by the light receiving element and a digital output value indicating an output pixel value in the scan process. The second scan image obtained after the input / output characteristic adjustment process for adjusting the input / output characteristics so that a part of the output range in the input / output characteristics before the input / output characteristic adjustment process becomes a wider output range. In the frequency distribution related to the pixel values of a plurality of pixels in the whole, a value between the newly obtained pixel value of the extra-manuscript region peak and the newly obtained pixel value of the manuscript region peak is set as a threshold value, and the threshold value is used as the threshold value. Based on the method of discriminating between the original area and the outer area of the original, the scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates of the original area. In discriminating between the manuscript region and the outer region of the manuscript based on the positions of the four orthogonal points where the edge candidates are orthogonal to each other, it is determined that at least one of the four orthogonal points is not detected. If so, the four orthogonal points are connected by a straight line based on the positional relationship and the geometrical relationship of the detection positions of the remaining orthogonal points. The positions of the at least one orthogonal point are estimated so that the closed region to be formed has a rectangular shape, the closed region is discriminated as the manuscript region, and the region other than the closed region is discriminated as the outer region of the manuscript. The scan image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates in the original area, and the length of the extracted edge candidates is longer than a predetermined value. The straight line portion is specified as a regular edge portion of the original document area, a portion other than the straight line portion is specified as a false detection portion, and complementary processing based on the regular edge portion is performed to correct the false detection portion. Then, one of the methods of discriminating the closed region formed by the corrected edge candidate as the document region and determining the region other than the closed region as the outside region of the document. a-1) In the frequency distribution relating to the pixel values of a plurality of pixels in the entire scanned image, there is a step of determining whether or not the distance between the extra-manuscript region peak and the manuscript region peak is larger than a certain degree. Then, in step b), b-1) the first determination method is the optimum method on condition that it is determined that the distance between the out-of-manuscript region peak and the manuscript region peak is larger than the certain degree. The second determination method is determined as the optimum method on condition that it is determined that the distance between the extra-manuscript region peak and the manuscript region peak is smaller than a certain degree. It is characterized by having a step to perform.
The invention of claim 32 is based on a) a step of analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen, and b) an analysis result in the step a). Based on this, the most suitable for the scanned image is among a plurality of discrimination methods for discriminating between a document area in which the document exists and an area outside the document in which the document does not exist in the image. A program for executing a step of determining the optimum method, which is a method, and a step of c) discriminating between the manuscript area and the non-manuscript area using the optimum method determined in step b). The plurality of discrimination methods include a first discrimination method and a fourth discrimination method, and the first discrimination method includes a plurality of discrimination methods existing in a frequency distribution relating to pixel values of a plurality of pixels in the entire scanned image. Of the peaks, the peak with the smallest pixel value at the peak position is estimated as the peak outside the document, which is the peak corresponding to the outside region, and the remaining peak is the peak corresponding to the outside region, which is the original region. In addition to estimating as a peak, a value between the pixel value of the extra-manuscript region peak and the pixel value of the manuscript region peak is used as a threshold, and the manuscript region and the extra-manuscript region are discriminated based on the threshold. In the fourth discrimination method, edge candidates in the original region are extracted by performing binarization processing and edge extraction processing based on a provisional threshold on the scanned image, and the edge candidates are orthogonal to each other at four points. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the original region from the outer region based on the position of the orthogonal point, the remaining orthogonal points Based on the positional relationship and the geometrical relationship of the detection positions of the above four points, the positions of the at least one orthogonal point are set so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape. A method of estimating, determining the closed area as the original area, and determining an area other than the closed area as the outer area of the original, and binarization processing and edge extraction based on a provisional threshold value for the scanned image. Processing is performed to extract edge candidates of the manuscript area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value is specified as a regular edge portion of the manuscript area, and other than the straight line portion. A portion is specified as a false detection portion, complement processing based on the regular edge portion is performed to correct the false detection portion, and the corrected edge candidate is formed. It is one of the methods of discriminating the closed area to be processed as the original area and determining the area other than the closed area as the outer area of the document. In the step a), a-1) a plurality of the scanned images. The step b) includes a step of determining whether or not a gradation of a predetermined degree or more exists in the entire scanned image based on the spatial distribution of the pixel values of the pixels of the above, and the step b) is b-1) the scan. The step of determining the fourth discrimination method as the optimum method on condition that the gradation of the predetermined degree or more exists in the entire image, and b-2) The gradation of the predetermined degree or more does not exist in the entire scanned image. The present invention is characterized in that it has a step of determining the first determination method as the optimum method, provided that
The invention of claim 33 is based on a) a step of analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen, and b) an analysis result in the step a). Based on this, the most suitable for the scanned image is among a plurality of discrimination methods for discriminating between a document area in which the document exists and an area outside the document in which the document does not exist in the image. A program for executing a step of determining the optimum method, which is a method, and a step of c) discriminating between the manuscript area and the non-manuscript area using the optimum method determined in step b). The plurality of discrimination methods include a first discrimination method and a fourth discrimination method, and the first discrimination method includes a plurality of discrimination methods existing in a frequency distribution relating to pixel values of a plurality of pixels in the entire scanned image. Of the peaks, the peak with the smallest pixel value at the peak position is estimated as the peak outside the document, which is the peak corresponding to the outside region, and the remaining peak is the peak corresponding to the outside region, which is the original region. In addition to estimating as a peak, a value between the pixel value of the extra-manuscript region peak and the pixel value of the manuscript region peak is used as a threshold, and the original region and the extra-manuscript region are discriminated based on the threshold. In the fourth discrimination method, edge candidates in the original region are extracted by performing binarization processing and edge extraction processing based on a provisional threshold on the scanned image, and the edge candidates are orthogonal to each other at four points. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the original region from the outer region based on the position of the orthogonal point, the remaining orthogonal points Based on the positional relationship and the geometrical relationship of the detection positions of the above four points, the positions of the at least one orthogonal point are set so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape. A method of estimating, determining the closed area as the original area, and determining an area other than the closed area as the outer area of the original, and binarization processing and edge extraction based on a provisional threshold value for the scanned image. Processing is performed to extract edge candidates of the manuscript area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value is specified as a regular edge portion of the manuscript area, and other than the straight line portion. The portion is specified as a false detection portion, the false detection portion is corrected by performing complementary processing based on the regular edge portion, and the corrected edge candidate is formed. It is one of the methods of discriminating the closed area to be processed as the original area and determining the area other than the closed area as the outer area of the original, and the step a) is a-1) the entire scanned image. In the frequency distribution related to the pixel values of a plurality of pixels, a step of determining whether or not the distance between the extra-manuscript region peak and the manuscript region peak is larger than a certain degree, and a-2) a plurality of scan images. The step b) includes a step of determining whether or not a gradation of a predetermined degree or more exists in the entire scanned image based on the spatial distribution of the pixel values of the pixels, and the step b) is b-1) the frequency. The step of determining the first determination method as the optimum method on condition that the distance between the out-of-manuscript region peak and the original document region peak is determined to be larger than a certain degree in the distribution, and b- 2) It is characterized by having a step of determining the fourth discrimination method as the optimum method on condition that a gradation of a predetermined degree or more is present in the entire scanned image.
The invention of claim 34 is based on a) a step of analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen, and b) an analysis result in the step a). Based on this, the most suitable for the scanned image is among a plurality of discrimination methods for discriminating between a document area in which the document exists and an area outside the document in which the document does not exist in the image. A program for executing a step of determining the optimum method, which is a method, and a step of c) discriminating between the manuscript area and the non-manuscript area using the optimum method determined in step b). The plurality of discrimination methods include a second discrimination method and a fourth discrimination method, and the second discrimination method shows an analog input value and an output pixel value indicating the amount of light received by the light receiving element in the scan process. It is an input / output characteristic adjustment process that adjusts the input / output characteristics with the digital output value, and is entered so that a part of the output range in the input / output characteristics before the input / output characteristic adjustment process becomes a wider output range. In the frequency distribution related to the pixel values of a plurality of pixels in the entire second scan image obtained after the input / output characteristic adjustment process for adjusting the output characteristics, the pixel value at the peak position is the smallest among the plurality of peaks. The peak is determined as the extra-manuscript region peak which is the peak corresponding to the extra-manuscript region, the residual peak is determined as the original region peak which is the peak corresponding to the extra-manuscript region, and the pixels of the extra-manuscript region peak are determined. A value between the value and the pixel value of the original document area peak is set as a threshold value, and the original document area and the outer document area are discriminated based on the threshold value. The fourth discriminating method is a method for discriminating the scanned image. On the other hand, edge candidates of the manuscript area are extracted by performing binarization processing and edge extraction processing based on a provisional threshold value, and the manuscript area and the manuscript are based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing from the outer region, it is based on the positional relationship and the geometric relationship of the detection positions of the remaining orthogonal points. Therefore, the positions of the at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is determined as the document region. , A method for discriminating an area other than the closed area as the extra document area, and binarization processing and edge extraction based on a provisional threshold value for the scanned image. Processing is performed to extract edge candidates of the manuscript area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value is specified as a regular edge portion of the manuscript area, and other than the straight line portion. A portion is specified as a false detection portion, complementary processing based on the regular edge portion is performed to correct the false detection portion, and a closed region formed by the corrected edge candidate is determined as the original document region. It is one of the methods for discriminating an area other than the closed area as the area outside the original document, and the step a) is a-1) based on the spatial distribution of the pixel values of a plurality of pixels in the scanned image. The step b) includes a step of determining whether or not a gradation of a predetermined degree or more exists in the entire scanned image, and b-1) the gradation of the predetermined degree or more exists in the entire scanned image. The second discrimination method is optimized on the condition that the step of determining the fourth discrimination method as the optimum method and b-2) that there is no gradation of the predetermined degree or more in the entire scanned image. It is characterized by having a step to be determined as a method.

According to the inventions of claims 1 to 34 , the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen is analyzed, and based on the analysis result, the image is included in the image. The optimum method, which is the most suitable method for the scanned image, is determined from a plurality of discrimination methods for discriminating between the document area and the area outside the document. Then, the manuscript area and the non-manuscript area are discriminated by using the determined optimum method. Therefore, in the scanned image of the document placed on the platen, the document area and the area outside the document can be more accurately discriminated.

It is a figure which shows the appearance of the image forming apparatus (MFP). It is a figure which shows the functional block of the MFP. It is a flowchart which shows the operation of the MFP. It is a figure which shows the document cover etc. in the fully opened state. It is a figure which shows the scanned image. It is a figure which shows the histogram. It is a figure which shows the scanned image. It is a figure which shows the histogram. It is a figure explaining about gain adjustment. It is a figure which shows the histogram about the pixel value after gain adjustment. It is a figure which shows the document cover etc. in a half-open state. It is a figure which shows the scan image which has the gradation of the main scanning direction. It is a figure which shows the histogram about the whole scan image which has the gradation of the main scanning direction. It is a figure which shows the division area and the histogram for each division area. It is a figure which shows the histogram for each division area. It is a figure which shows the scan image which has the gradation in an oblique direction. It is a figure which shows the histogram about the whole scan image which has a gradation in an oblique direction. It is a figure which shows the orthogonal point and the like detected in the scanned image. It is a figure explaining the correction of an orthogonal point and the like. It is a figure which shows the extracted edge candidate and the like. It is a figure explaining the correction of the erroneous detection part. It is a figure explaining the correction of the erroneous detection part which concerns on the modification. It is a figure which shows the estimation result of a manuscript area and an area outside a manuscript. It is a figure which shows the scan image which has the gradation of the sub-scanning direction. It is a figure which shows the histogram. It is a figure which shows the scanned image which has the gradation in an oblique direction, and the division area.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.

<1-1. Configuration overview>
FIG. 1 is a diagram showing the appearance of the image forming apparatus 10. Here, as the image forming apparatus 10, the MFP (Multi-Functional Peripheral) 10 is illustrated.

As shown in FIGS. 4 and 11, the MFP 10 includes a document base 140 on which a document is placed and a document cover 110. The document cover 110 has a rotary opening / closing mechanism (rotary opening / closing type). Specifically, the document cover 110 is attached to the document base 140 by a hinge portion 130 for opening and closing, and is provided so as to be rotatable around the hinge portion 130 as a rotation axis. Further, here, the hinge portion 130 is formed along the X direction at one end in the depth direction (Y direction) of the platen 140 (here, the end portion on the back side (opposite the installation position of the operation panel portion 6c)). Have been placed.

FIG. 2 is a diagram showing a functional block of the MFP 10.

The MFP 10 is a device (also referred to as a multifunction device) having a scanning function, a copying function, a facsimile function, a box storage function, and the like. Specifically, as shown in the functional block diagram of FIG. 2, the MFP 10 includes an image reading unit 2, a print output unit 3, a communication unit 4, a storage unit 5, an operation unit 6, a controller (control unit) 9, and the like. By operating each of these parts in a complex manner, various functions are realized. The MFP 10 is also referred to as an image processing device or an image reading device.

The image scanning unit 2 optically scans (that is, scans) a document placed on the document mounting glass surface of the document table 140, and scans an image (also referred to as a scanned image) of the document on the document table 140. ) (Detailedly image data) is a processing unit. The image reading unit 2 is also referred to as a scanning unit. In the MFP 10, the Y direction (depth direction) (vertical direction) on the document mounting glass surface of the platen 140 is the main scanning direction in the scanning operation, and the X direction (X direction) on the document mounting glass surface. The (horizontal direction) (horizontal direction) is the sub-scanning direction in the scanning operation (see FIGS. 1, 4, etc.).

The print output unit 3 is an output unit that prints and outputs an image on various media such as paper based on data related to a print target.

The communication unit 4 is a processing unit capable of performing facsimile communication via a public line or the like. Further, the communication unit 4 can also perform network communication via the network. In this network communication, for example, various protocols such as TCP / IP (Transmission Control Protocol / Internet Protocol) are used. By using the network communication, the MFP 10 can send and receive various data in cooperation with a desired destination. The communication unit 4 has a transmission unit 4a for transmitting various data and a reception unit 4b for receiving various data.

The storage unit 5 is composed of a storage device such as a hard disk drive (HDD).

The operation unit 6 includes an operation input unit 6a that receives an operation input to the MFP 10 and a display unit 6b that displays and outputs various information.

The MFP 10 is provided with a substantially plate-shaped operation panel portion 6c (see FIG. 1). Further, the operation panel unit 6c has a touch panel 25 (see FIG. 1) on the front side thereof. The touch panel 25 functions as a part of the operation input unit 6a and also as a part of the display unit 6b. The touch panel 25 is configured by embedding various sensors and the like in a liquid crystal display panel, and can display various information and receive various operation inputs from an operator (user). For example, an operation screen is displayed on the touch panel 25.

The controller (control unit) 9 is a control device that is built in the MFP 10 and controls the MFP 10 in an integrated manner. The controller 9 is configured as a computer system including a CPU and various semiconductor memories (RAM and ROM). The controller 9 realizes various processing units by executing a predetermined software program (hereinafter, also simply referred to as a program) stored in a ROM (for example, EEPROM (registered trademark)) in the CPU. The program (specifically, a program module group) may be recorded on a portable recording medium such as a USB memory and installed on the MFP 10 via the recording medium. Alternatively, the program may be downloaded via a network or the like and installed in the MFP 10.

Specifically, as shown in FIG. 2, the controller 9 executes the program to execute the communication control unit 11, the input control unit 12, the display control unit 13, the analysis unit 14, the area determination unit 17, and the out-of-manuscript erasure unit. Various processing units including 18 are realized.

The communication control unit 11 is a processing unit that controls communication operations with other devices in cooperation with the communication unit 4 and the like.

The input control unit 12 is a control unit that controls an operation of receiving an operation input from a user with respect to the operation input unit 6a (touch panel 25 or the like).

The display control unit 13 is a processing unit that controls the display operation on the display unit 6b (touch panel 25 or the like).

The analysis unit 14 is a processing unit that executes a process of analyzing the distribution state of the pixel values of a plurality of pixels in the scanned image 300 regarding the document placed on the platen 140. For example, the analysis unit 14 has a gradation (gradation change) of a predetermined degree or more in the entire scanned image 300 based on the spatial distribution of the pixel values of a plurality of pixels in the scanned image 300 (see FIG. 5 and the like). Determine whether or not to do so. Further, when the scanned image 300 has a gradation of the predetermined degree or more, the analysis unit 14 further executes a process of analyzing (determining) the change direction of the gradation of the predetermined degree or more based on the spatial distribution situation. To do. Further, the analysis unit 14 analyzes (analyzes) the frequency distribution of the pixel values of the plurality of pixels based on the histogram (frequency distribution) of the pixel values of the plurality of pixels in the entire scanned image 300.

The area determination unit 17 is a processing unit that executes an area determination process for discriminating between the original area (the area where the original exists) and the outer area (the area where the original does not exist) in the scanned image 400 (see FIG. 5 and the like). .. The area discrimination unit 17 is a plurality of discrimination methods (plural types of discrimination methods) for discriminating between the original region and the non-manuscript region in the image based on the analysis result (analysis result of the distribution status) by the analysis unit 14. From among them, the most suitable method (optimal method) for the scanned image 300 is determined. In other words, a plurality of discrimination methods can be switched according to the analysis result. Then, the area discrimination unit 17 discriminates between the original area and the non-manuscript area in the scanned image 400 by using the determined optimum method. In addition, the area determination unit 17 also executes a threshold value calculation process for calculating a threshold value (extraction threshold value) TH for extracting edge candidates in the document area.

The out-of-manuscript erasing unit 18 is a processing unit that executes an out-of-manuscript erasing process for erasing the out-of-manuscript area (the area determined as the out-of-manuscript area) from the scanned image 400 based on the discrimination result between the out-of-manuscript area and the out-of-manuscript area. Is. Specifically, the extra-manuscript erasing unit 18 replaces (corrects) the pixel value of the region determined as the extra-manuscript region with white data (specifically, the pixel value "255") (executes the white replacement processing). ..

<1-2. Operation>
FIG. 3 is a flowchart showing the operation of the MFP 10.

The MFP 10 analyzes the distribution of pixel values of a plurality of pixels in the scanned image 300 regarding the document placed on the platen 140 (steps S13 to S15). Then, the MFP 10 is the most suitable method (most suitable for the scanned image 300) from among a plurality of discrimination methods (steps S16 to S19) for discriminating between the document region and the region outside the document based on the analysis result of the distribution situation. Method) is determined, and the original area and the non-manuscript area are discriminated by using the determined optimum method. When the document area and the document area are determined, the MFP 10 erases the document area from the scanned image 400 (step S20). Then, a print output process or the like of the erasing result of the area outside the document (erasing result outside the document) is executed.

Hereinafter, after the outline of the process of FIG. 3 will be described, the contents of each discrimination method (steps S16 to S19) will be described.

First, when executing a job involving scanning (for example, a copy job), the user opens the document cover 110 (also referred to as a platen cover) of the MFP 10 and opens the document table 140 glass surface (also referred to as platen glass) for placing documents. Place the manuscript in). Then, the user gives the execution instruction of the job to the MFP 10.

In response to the user giving an execution instruction, the MFP 10 starts the process of FIG.

In step S11, the MFP 10 executes a pre-scan (a scanning process for generating a preparatory image for the off-document erasing process). Specifically, the MFP 10 scans the original mounted on the platen 140 by performing a scanning operation over the entire surface of the platen 140 for placing the original, and scans the image 300 (see FIG. 5 and the like). ) Is generated. The scanned image 300 is generated as a preparatory image (an image for analyzing the distribution state, etc.) of the non-manuscript erasing process.

Specifically, in response to the user giving an execution instruction, light is emitted from the light source 150 (see FIG. 4) arranged below the glass surface for placing the document, and the emitted light is emitted from the glass surface for placing the document. Is transparent. Then, a part of the emitted light hits the document on the platen 140, is reflected, passes through the platen 140 again, is incident on the light receiving portion (not shown), and is detected by the light receiving portion. After that, a scanned image 300 (see FIG. 5 and the like) is generated based on the intensity of light detected by the light receiving unit and the like. In the scanned image 300, the pixel value of the pixel (for example, the pixel in the document area) of the portion where relatively strong light is detected is relatively large (bright), and the portion (or light) where relatively weak light is detected is detected. The pixel value of the pixel (the undetected part) is relatively small (dark).

When the scanned image 300 is generated, the MFP 10 tentatively calculates the extraction threshold TH for extracting edge candidates in the document region based on the scanned image 300, and also tentatively calculates the tentatively calculated extraction threshold value TH. The original area and the outer area of the original are estimated based on the threshold value TH (step S12). As will be described later, the result obtained in step S12 is used in the analysis of the distribution of the pixel values of the plurality of pixels in the scanned image 300 (steps S13 to S15), and is tentatively calculated. The extraction threshold TH is used in some discriminating techniques.

Specifically, the MFP 10 generates a histogram (frequency distribution) for the pixel values of a plurality of pixels in the entire scanned image 300, and sets an extraction threshold TH for extracting edge candidates in the document region based on the histogram. Temporarily calculate (temporarily set). For example, when two peaks (peaks whose peak value (maximum value) has a frequency of a certain degree or more) exist in the histogram (see FIG. 6 and the like), the peak corresponding to the manuscript area and the peak outside the manuscript area A provisional extraction threshold TH is set (temporarily set) between the peak and the peak. Alternatively, when a single peak exists in the histogram (see FIG. 13), a provisional extraction threshold TH is set (temporarily set) at the base portion (for example, the base portion on the left side) of the single peak. ..

Then, based on the provisionally calculated (temporarily set) extraction threshold value TH, binarization processing is executed for each pixel in the scanned image 300, and edge candidates of the original region are extracted based on the processing result. Edge extraction processing (differential filter processing, etc.) is executed. After that, in the scanned image 300, the closed region formed by the extracted edge candidates is estimated as the original region, and the region other than the closed region is estimated as the outer region of the original.

When the document area and the area outside the document are estimated (step S12), the MFP 10 analyzes (analyzes) the distribution of the pixel values of the plurality of pixels in the scanned image 300 (steps S13 to S15). Then, based on the analysis result (analysis result), the optimum method (optimum) is selected from a plurality of different discrimination methods (plural types of discrimination methods) (steps S16 to S17) for discriminating between the manuscript area and the non-manuscript area. Discrimination method) is determined. In other words, from the plurality of discrimination methods, a different discrimination method is determined (adopted) as the optimum method according to the analysis result of the distribution situation.

First, in step S13, it is determined whether or not there is a gradation of a predetermined degree or more in the entire scanned image 300 based on the spatial distribution of the pixel values of the plurality of pixels in the scanned image 300. In other words, the presence / absence determination process of the gradation of a predetermined degree or more in the entire scanned image 300 is executed.

Specifically, the MFP 10 extracts at least one (here, both) of the vertical pixel row (the pixel row extending in the vertical direction) and the horizontal pixel row (the pixel row extending in the horizontal direction) from the scanned image 300. , Analyze (analyze) the distribution status (spatial distribution status) of the pixel values of each pixel sequence. Then, the MFP 10 determines the presence / absence of a predetermined change characteristic (described below) with respect to the at least one pixel string, and based on the determination result of the presence / absence of the predetermined change characteristic, the entire scanned image 300 has a gradation of a predetermined degree or more. Determines if is present.

Specifically, the MFP 10 extracts both a specific vertical pixel sequence and a specific horizontal pixel sequence as the pixel strings of interest in the scanned image 300, and determines the presence or absence of a predetermined change characteristic in each pixel sequence of interest. To do. More specifically, in the MFP 10, the pixel values of the plurality of pixels in the extension direction of the pixel sequence of interest gradually change (increase or decrease) with respect to each pixel sequence of interest, and the width of change in the pixel values of the plurality of pixels. It is determined whether or not the (variation width) has a predetermined change characteristic exceeding a predetermined degree (for example, the pixel value “50”). In short, the MFP 10 pays attention to both the vertical pixel row and the horizontal pixel row, and whether at least one of the vertical pixel row and the horizontal pixel row has a gradation of a predetermined degree or more. Whether or not it is determined.

More specifically, a specific vertical pixel row among a plurality of vertical pixel trains arranged in the horizontal direction in the scanned image 300 (a vertical pixel row including a specific vertical pixel row in the outer region of the original document). Is extracted as a pixel sequence of interest (a pixel array in the vertical direction of attention). For example, when all or a part of the left end (and / or right end) vertical pixel row in the scanned image 300 is included in the outer region of the document among the plurality of vertical pixel rows, the leftmost vertical pixel row is included. It is extracted as a pixel sequence of interest (a pixel array in the vertical direction of attention). Then, the spatial distribution of the pixel values of the plurality of pixels in the attention vertical pixel array is analyzed, and the presence or absence of a predetermined change characteristic is determined for the plurality of pixels in the attention vertical pixel array. In other words, it is determined whether or not there is a gradation of a predetermined degree or more in the vertical pixel sequence.

Further, a specific pixel string (horizontal pixel string including a specific horizontal pixel string in the region outside the original document) among a plurality of horizontal pixel strings arranged in the vertical direction in the scanned image 300 is a pixel string of interest (a pixel string in the horizontal direction). It is extracted as a pixel sequence in the horizontal direction of interest). For example, when all or a part of the upper end (and / or lower end) horizontal pixel row in the scanned image 300 is included in the outer region of the document among the plurality of horizontal pixel rows, the upper end horizontal pixel row is included. It is extracted as a pixel sequence of interest (horizontal pixel array of interest). Then, the spatial distribution of the pixel values of the plurality of pixels in the attention horizontal pixel row is analyzed, and the presence or absence of a predetermined change characteristic is determined with respect to the plurality of pixels in the attention horizontal pixel row. In other words, it is determined whether or not there is a gradation of a predetermined degree or more in the horizontal pixel sequence.

Then, different discrimination methods are determined (adopted) as the optimum method according to the determination result (analysis result) of whether or not the entire scanned image 300 has a gradation of a predetermined degree or more. In other words, different discrimination methods are determined (adopted) as the optimum method depending on whether the entire scanned image 300 has a gradation of a predetermined degree or more and the entire scanned image 300 does not have a gradation of a predetermined degree or more. ..

For example, when at least one of the vertical pixel row (attention vertical pixel row) and the horizontal pixel row (attention horizontal pixel row) has a gradation of a predetermined degree or more, the entire scanned image 300 has a gradation of a predetermined degree or more. Is determined in step S13. Then, the process proceeds to step S15 or later, and either the third discrimination method (step S18) or the fourth discrimination method (step S19) is determined as the optimum method.

On the other hand, when the gradation of the predetermined degree or more does not exist in both the vertical pixel row and the horizontal pixel row, it is determined in step S13 that the gradation of the predetermined degree or more does not exist in the entire scanned image 300. Then, the process proceeds to step S14 and subsequent steps, and either the first discrimination method (step S16) or the second discrimination method (step S17) is determined as the optimum method.

In step S14, the frequency distribution status of the pixel values of the plurality of pixels is analyzed (analyzed) based on the frequency distribution relating to the pixel values of the plurality of pixels in the entire scanned image 300 (the frequency distribution status analysis process is executed). Will be).

Specifically, a condition that the distance between the original region peak and the outer region peak is larger than a certain degree in the histogram (frequency distribution) generated in step S12 with respect to the pixel values of a plurality of pixels in the entire scanned image 300. Is determined whether or not is satisfied. In other words, it is determined whether or not the condition that the difference between the pixel value of the original region peak and the pixel value of the extra-manuscript region peak is larger than a predetermined value (for example, the value "50") is satisfied in the histogram. Will be done. The document area peak is a peak corresponding to the document area in the histogram, and the document area is a peak corresponding to the document area in the histogram.

Then, different discrimination methods are determined as the optimum method depending on whether or not the condition that the distance between the peaks is larger than a certain degree is satisfied. In other words, there are different discrimination methods depending on whether the condition that the distance between each peak is larger than a certain degree is satisfied and the condition that the distance between each peak is larger than a certain degree is not satisfied. Determined as the optimal method.

For example, when two or more peaks are detected in the histogram and the distance between the original region peak and the extra-manuscript region peak is larger than a certain degree, the distance between the original region peak and the extra-manuscript region peak is larger than a certain degree in the histogram. It is determined that the condition of largeness is satisfied, and the process proceeds to step S16. Then, the first discrimination method is determined (adopted) as the optimum method. In other words, the condition that there is no gradation of a predetermined degree or more in the entire scanned image 300 and that the distance between the original region peak and the outer region peak in the original region is larger than a certain degree in the histogram for the entire scanned image 300 is satisfied. If so, the manuscript area and the non-manuscript area are discriminated using the first discrimination method. Here, a method of discriminating between the original region and the non-manuscript region based on the extraction threshold TH (TH10) calculated based on the histogram of the pixel values of a plurality of pixels in the entire scanned image 300 (also referred to as “basic discrimination method”). ) Is used as the first discrimination method. The details of the first discrimination method (basic discrimination method) will be described later.

On the other hand, if the condition that the distance between the original region peak and the outer region peak is larger than a certain level is not satisfied in the histogram, the process proceeds to step S17, and the second discrimination method is determined (adopted) as the optimum method. Will be done. For example, if the original region peak and the extra-manuscript region peak are not specified (detected) in the histogram, the condition that the distance between the original region peak and the extra-manuscript region peak is larger than a certain level in the histogram is not satisfied. The discrimination method (step S17) of 2 is determined as the optimum method. Further, although the original region peak and the extra-manuscript region peak are detected, even if the distance between the original region peak and the extra-manuscript region peak is smaller than a certain degree, the original region peak and the extra-manuscript region peak are displayed in the histogram. The condition that the distance is larger than a certain degree is not satisfied, and the second discrimination method (step S17) is determined as the optimum method. In other words, the condition that there is no gradation of a predetermined degree or more in the entire scanned image 300 and that the distance between the original region peak and the outer region peak in the original region is larger than a certain degree in the histogram for the entire scanned image 300 is satisfied. If not, the manuscript area and the out-of-manuscript area are discriminated by using the second discrimination method. Here, the input / output characteristic adjustment processing is executed for the pixel values of a plurality of pixels in the scanned image 300, and the calculation is performed based on a new histogram of the pixel values obtained based on the input / output characteristics after the input / output characteristic adjustment processing. A method of specifying the original area and the outer area of the original using the extraction threshold TH (TH20) (also referred to as “input / output characteristic adjustment method”) is used as the second discrimination method. The details of the second discrimination method (input / output characteristic adjustment method) will be described later.

Further, as described above, when it is determined in step S13 that at least one of the vertical pixel array and the horizontal pixel array has a gradation of a predetermined degree or more, the process proceeds to step S15.

In step S15, the change direction of the gradation of a predetermined degree or more existing in the scan image 300 is further analyzed (determined) based on the spatial distribution of the pixel values of the plurality of pixels in the scan image 300. In other words, the direction determination process for determining the change direction of the gradation of a predetermined degree or more in the entire scanned image is executed. Specifically, in the scanned image 300, the change direction of the gradation of the predetermined degree or more is determined based on the determination result of the presence / absence of the predetermined change characteristic in both the vertical pixel row and the horizontal pixel row.

Then, different discrimination methods are determined (adopted) as the optimum method depending on whether the change direction is the vertical direction or the horizontal direction in the scanned image 300 or the oblique direction in the scanned image 300. In other words, different discrimination methods are determined as the optimum method depending on whether the change direction is the vertical direction or the horizontal direction and the change direction is the diagonal direction.

For example, when it is determined in step S15 that the change direction is the vertical direction or the horizontal direction, the process proceeds to step S18, and the third determination method is determined (adopted) as the optimum method. In other words, when there is a gradation of a predetermined degree or more in the entire scanned image 300, and the change direction of the gradation of the predetermined degree or more is the vertical direction or the horizontal direction, the original area and the original are used by the third discrimination method. It is determined to be the outer area. Here, a method of dividing the scanned image 300 into a plurality of divided regions 380 and discriminating between the original region and the outer region of the original in each divided region based on the extraction threshold value TH calculated for each divided region (“region division method”). ”) Is used as the third discrimination method. The details of the third discrimination method (region division method) will be described later.

On the other hand, when it is determined in step S15 that the change direction of the gradation of the predetermined degree or more is an oblique direction, the process proceeds to step S19, and the fourth discrimination method different from the third discrimination method is optimal. Determined (adopted) as a method. In other words, when there is a gradation of a predetermined degree or more in the entire scanned image 300 and the change direction of the gradation of the predetermined degree or more is an oblique direction, the original region and the outer region of the original are determined by using the fourth discrimination method. Is determined. Here, edge candidates in the document region are extracted from the scanned image 300, and the positions of the undetected orthogonal points among the four orthogonal points of the edge candidates are estimated based on the detection positions of the remaining orthogonal points. A method of discriminating between a manuscript area and a region outside the manuscript based on four orthogonal points (also referred to as an “orthogonal point estimation method”) is used as the fourth discriminating method. The details of the fourth discrimination method (orthogonal point estimation method) will be described later.

In this way, the distribution of pixel values of a plurality of pixels in the scanned image 300 is analyzed (steps S13 to S15), and based on the analysis results, the most suitable discrimination methods (optimal methods) for the scanned image 300 are mutually It is determined from four different types of discrimination methods (steps S16 to S19). Then, the manuscript area and the non-manuscript area are discriminated by using the determined discrimination method.

After that, when the original area and the outer area are determined, the process proceeds to step S20, and the extra-manuscript erasing process for erasing the outer area from the scanned image 400 (described later) is executed. Specifically, the pixel value of the region determined as the region outside the original document is replaced (corrected) with white data (specifically, the pixel value "255"). Here, the discrimination method determined as the optimum method for the scanned image 300 is also the most suitable discrimination method for the scan image 400 (described later), and the discrimination method determined as the optimum method for the scan image 300 is used. In the scanned image 400, the document area and the area outside the document are discriminated. The scan image 400 is a scan image generated in the main scan described later, and is a scan image captured in the same situation as the scan image 300 (scan image generated by the pre-scan).

Next, the contents of the four types of discrimination methods (steps S16 to S19) will be described below.

<About the first discrimination method (basic discrimination method)>
As will be described later, when the scanned image 300 (310) as shown in FIG. 5 is generated, the first discrimination method (basic discrimination method) among the plurality of discrimination methods is determined as the optimum method.

Here, after the original is placed on the glass surface for placing the original, the user sets the original cover 110 in the fully open state (for example, the original cover 110 is opened to the extent orthogonal to the original base 140) ( (See also FIG. 4), and a job execution instruction is given. In response to the execution instruction, the MFP 10 executes a pre-scan (step S11) to generate a scanned image 300.

Here, when the document cover 110 is in the fully open state, among the light emitted from the light source 150 and transmitted through the glass surface for placing the document, the light that does not hit the document on the platen 140 is incident on the light receiving portion. It is not detected by the light receiving unit. In this case, in step S11, the scanned image 310 as shown in FIG. 5 is generated based on the intensity of the light detected by the light receiving unit and the like. In the scanned image 310, since the light reflected on the document (relatively strong light) is detected, the pixel values of the pixels in the document area are relatively large (bright), and the pixel values of the pixels in the area outside the document are relatively large (bright). Relatively small (dark).

Then, in the scanned image 310 generated by the pre-scan, the original area and the outer area of the original are estimated (step S12). Specifically, the MFP 10 generates a histogram for the pixel values of a plurality of pixels in the entire scanned image 310, and tentatively calculates an extraction threshold TH10 for extracting edge candidates in the original region based on the histogram. To do. Then, based on the tentatively calculated extraction threshold value TH10, the original region and the outer region of the original are estimated in the scanned image 310. FIG. 6 is a diagram showing a histogram H10 regarding the pixel value of each pixel in the scanned image 310 (FIG. 5).

When the document cover 110 is in the fully open state, in the histogram H10 generated for the scanned image 310, as shown in FIG. 6, a peak having a relatively large pixel value (luminance value) B and a relatively small luminance value B Two peaks are formed with a peak having. Specifically, the peak existing on the right side in the histogram H10 indicates that there are relatively many pixels in which relatively strong light is detected, and the peak existing on the left side in the histogram H10 is detected by relatively weak light. It shows that the number of pixels is relatively large. Then, the MFP 10 detects the leftmost peak (the peak having the smallest peak position) among the plurality of peaks (here, two peaks) existing in the histogram H10 as the extra-manuscript region peak. Further, the MFP 10 detects the remaining peaks other than the peaks in the outer region of the original document as the original region peaks among the plurality of peaks (two peaks in this case) existing in the histogram H10.

Then, based on the two peaks in the histogram H10, an extraction threshold value TH10 for extracting edge candidates in the document region is tentatively calculated. For example, between the manuscript region peak and the out-of-manuscript region peak, the absolute value of the slope of the curve of the out-of-manuscript region peak from the middle to the base of the out-of-manuscript region peak is The extraction threshold TH10 is tentatively set (temporarily set) at a position (position reduced to a predetermined degree).

Then, based on the provisional extraction threshold TH10, the original region and the extra-manuscript region are estimated in the scanned image 310.

Specifically, in the scanned image 310, a pixel group having a pixel value larger than the extraction threshold TH10 provisionally calculated by the histogram H10 (FIG. 6) is determined as a pixel group in the document region. Then, in the scan image 310, binarization processing is executed for each pixel in the scan image 310 based on the magnitude relationship with the provisional extraction threshold TH10. Further, an edge extraction process for extracting edge candidates in the original area is executed based on the processing result of the binarization process, and the edge candidates are extracted. Then, the closed region formed by the extracted edge candidates is estimated as the original region (temporary original region) in the scanned image 310, and the region other than the closed region is used as the outer region of the original (temporary outer region). Presumed.

When the document area and the area outside the document are estimated (step S12), the distribution of the pixel values of the plurality of pixels in the scanned image 300 is analyzed (steps S13 to S15). Then, the most suitable method (optimal method) for the scanned image 310 (410 (described later)) is determined from the four types of discrimination methods (steps S16 to S17) for discriminating between the document area and the area outside the document. Will be done.

First, in step S13, it is determined whether or not there is a gradation of a predetermined degree or more in the entire scanned image 310. Here, in the scanned image 310, it is determined that the attention vertical pixel row (for example, the leftmost vertical pixel row) does not have a gradation of a predetermined degree or more, and the attention horizontal pixel row (for example, the uppermost horizontal pixel row) is determined. It is determined that there is no gradation above a predetermined level even in the column). As described above, when there is no gradation of a predetermined degree or more in both the vertical pixel row and the horizontal pixel row, it is determined in step S13 that there is no gradation of a predetermined degree or more in the entire scanned image 300. To. Then, the process proceeds to step S14.

In step S14, based on the histogram H10 (see FIG. 6) generated in step S12 with respect to the pixel values of the plurality of pixels for the entire scanned image 310 (see FIG. 5), the original region peak and the extra-manuscript region peak are Whether or not the distance is larger than a certain degree is determined. Here, in the histogram H10, two or more peaks are detected, and the difference between the pixel value of the original region peak and the pixel value of the extra-manuscript region peak is larger than a predetermined value (for example, the value “50”). Therefore, in the histogram H10, it is determined that the condition that the distance between the document region peak (the peak on the right side) and the peak outside the document region (the peak on the left side) is larger than a certain degree is satisfied. Then, the process proceeds to step S16, and the first discrimination method (basic discrimination method) is determined (adopted) as the optimum method.

In the first discrimination method (basic discrimination method), the extraction threshold value TH10 tentatively calculated in step S12 is used as it is, and the manuscript area and the non-manuscript area are discriminated in the scanned image 400 (described later). To.

Specifically, first, the MFP 10 executes the main scan (scan processing for generating the (original) processing target image of the non-original erasing process), and scan image 400 (here, 410 (FIG. 5)). Is generated as an image for erasing outside the document (image to be processed by the erasing process outside the document) for erasing the area outside the document.

Then, in the scan image 410, the edge candidate of the document region is based on the extraction threshold value TH10 (extraction threshold value calculated based on the histogram H10 for the entire scan image 310) tentatively calculated (temporarily set) in step S12. Is extracted, and the manuscript area and the non-manuscript area are discriminated.

Specifically, in the scanned image 410, a pixel group having a pixel value larger than the extraction threshold TH10 is determined as a pixel group in the document region. Then, in the scan image 410, binarization processing is executed for each pixel in the scan image 410 based on the magnitude relationship with the extraction threshold TH10. Further, edge candidates of the document area are extracted based on the processing result of the binarization process. Then, the closed region formed by the extracted edge candidates is determined as the original region in the scanned image 410, and the region other than the closed region is determined as the outer region of the original.

When the document area and the area outside the document are discriminated by the scanned image 410, the first discriminating method (basic discriminating method) ends, and the process proceeds to step S20. Then, the area outside the document is erased from the scanned image 410 (step S20).

As described above, when the scan image 300 does not have a gradation of a predetermined degree or more and the condition that the distance between the document region peak and the document outer region peak is larger than a certain degree in the histogram is satisfied, the scan image 300 is satisfied. The original area and the outer area of the original are discriminated based on the extraction threshold value TH calculated based on the histogram of the pixel values of the plurality of pixels in the image.

<About the second discrimination method (input / output characteristic adjustment method)>
By the way, in the first discrimination method (basic discrimination method), the extraction threshold value TH tentatively set in step S12 is used as it is, and the document region and the region outside the document are discriminated. In other words, in the first discrimination method, the original region and the non-manuscript region are discriminated based on the histogram generated by using the pixel values of the plurality of pixels in the scanned image 300 as they are.

Here, when the pixel values of the plurality of pixels in the scanned image 300 are used as they are, the extraction threshold value TH may not be set (calculated) accurately.

Specifically, when the difference between the pixel value of the pixel in the original area and the pixel value of the pixel in the outer area of the original in the scanned image 300 is relatively small (in the peripheral area of the boundary between the original area and the outer area of the original), the above It may not be possible to accurately discriminate between the original area and the outer area using the basic discrimination method.

For example, a document with a relatively dark background (for example, a dark gray document) has a lower light reflectance than a document with a relatively light background (for example, a white document). Therefore, when a relatively dark background document is scanned, in the histogram H20 (H21) (see FIG. 8) generated for the scanned image 300 (320), the peak corresponding to the document area has a relatively small luminance value. Have. The peak corresponding to the out-of-manuscript region (the peak in the out-of-manuscript region) is largely shifted to the left side of the figure as compared with FIG. As a result, for example, as shown in FIG. 8, it becomes impossible to distinguish between the peaks in the outer region of the original document and the peaks in the original region, and a single pixel in the outer region of the original document and the pixels in the original document region coexist. A peak is formed. Further, even if two peaks, an extra-manuscript region peak and a manuscript region peak, are formed (when both peaks can be distinguished) as in the histogram H25 of FIG. 25, the two peaks are still present. It is formed in a relatively close position.

When a single peak is formed in the histogram (see FIG. 8), it is difficult to specify the position (pixel value) to be set as the extraction threshold TH. Further, even when two peaks are formed in the histogram (see FIG. 25), when the distance between the two peaks is relatively short, the extraction threshold value is set at a specific position in the foot portion of the peak outside the document. TH20 is not set accurately. As a result, the document area and the area outside the document may not be accurately discriminated (erroneous detection may occur).

In consideration of this point, when the scanned image 300 in which the difference between the pixel value of the document area and the pixel value of the outside document area is relatively small is generated, a second discrimination method different from the first discrimination method is used. It is used as an optimum method to discriminate between the original area and the non-original area. Specifically, as will be described later, when the condition that the distance between the original region peak and the extra-manuscript region peak is larger than a certain degree is not satisfied in the histogram regarding the scanned image 300 (320), a plurality of the scanned image 300 An accurate extraction threshold value TH is calculated by executing input / output characteristic adjustment processing with respect to the pixel value of the pixel to increase the distance between each peak in the histogram.

Here, it is assumed that the prescan is executed with the document cover 110 fully open (step S11), as in the case of the first discrimination method (basic discrimination method).

Then, after the scanned image 320 (321) (see FIG. 7) is generated in the pre-scan (step S11), the original area and the extra-manuscript area are estimated in the scanned image 321 (step S12).

Specifically, a histogram H20 (H21) (see FIG. 8 (or FIG. 25)) is generated for the pixel values of a plurality of pixels in the entire scanned image 321, and edge candidates of the document region are extracted based on the histogram H21. The extraction threshold TH20 (TH21) for this purpose is tentatively calculated (temporarily set).

For example, when the histogram H21 of FIG. 8 is generated, since two peaks are not detected, the absolute value of the slope of the curve is predetermined at a specific position of the histogram H21 (for example, in the middle portion of a single peak). The extraction threshold TH21 is tentatively set at the position (position reduced to the extent).

Further, when the histogram H21 of FIG. 25 is generated, since two peaks are detected, the extraction threshold TH21 is tentatively set between the peak outside the original region and the peak in the original region. After that, in the same manner as in the first discrimination method, the original region and the outer region of the original are estimated from the scanned image 321.

When the document area and the area outside the document are estimated (step S12), the distribution of pixel values of the plurality of pixels in the scanned image 321 is analyzed (steps S13 to S15).

Here, in step S13, it is determined that, similarly to the scanned image 310 (FIG. 5), there is no gradation of a predetermined degree or more in the scanned image 321 (FIG. 7). Then, the process proceeds to step S14.

In step S14, the distance between the document region peak and the document outside region peak is larger than a certain level in the histogram H21 generated in step S12 with respect to the pixel values of the plurality of pixels in the entire scanned image 321 (see FIG. 7). It is determined whether or not the condition of is satisfied.

For example, when the histogram H21 of FIG. 8 is generated, two or more peaks are not detected and the distance between the two peaks cannot be obtained. It is determined that the condition that the distance is larger than a certain degree is not satisfied.

Further, when the histogram H21 of FIG. 25 is generated, the distance between the two peaks is obtained, but it is determined in the histogram H21 that the distance between the original region peak and the outer region peak is smaller than a certain degree. Will be done. Then, in the histogram H21, it is determined that the condition that the distance between the document region peak and the document outer region peak is larger than a certain degree is not satisfied.

Then, the process proceeds to step S17, and the second discrimination method (input / output characteristic adjustment method) is determined (adopted) as the most suitable optimum method for the scanned image 321 (420 (described later)).

In the second discrimination method (input / output characteristic adjustment method), the extraction threshold value TH20 (TH21) tentatively calculated in step S12 is not used, and the input / output characteristics are adjusted to create a new extraction threshold value TH20 (TH20). TH25) (see also FIG. 9) is calculated. Then, the original region and the outer region of the original are discriminated based on the new extraction threshold TH25.

Specifically, the MFP 10 re-executes the pre-scan (in other words, executes the second pre-scan) to newly generate the scanned image 320 (322) (not shown). The scanned image 322 is an image obtained by A / D conversion based on the input / output characteristics after the input / output characteristic adjustment processing described later. Then, the MFP 10 calculates the new extraction threshold TH25 based on the scanned image 322.

Specifically, the MFP 10 performs input / output characteristic adjustment processing (so-called gain adjustment processing) for adjusting input / output characteristics for determining pixel values of a plurality of pixels in the scanned image 321 prior to the second prescan. Run. More specifically, as shown in FIG. 9, a part range R1 of the output range in the input / output characteristics before the input / output characteristic adjustment processing (pixel value having a frequency of a certain degree or more in the histogram H21 regarding the scanned image 321) is used. The input / output characteristics are adjusted so that the included partial range (see also FIG. 8) is the wider output range R2 (here, the entire range). That is, by adjusting the input / output characteristics, the correspondence between the input value (the amount of charge (analog value) generated by the photoelectric conversion action in each pixel (light receiving element) of the CCD) and the output value (pixel value (digital value)). The relationship is changed and the range of output values corresponding to the input values of the specific range W1 is increased from the range R1 to R2 (see FIG. 9). As a result, the input value in the specific range W1 is quantized more appropriately.

Then, the second prescan is executed. In the second pre-scan, the pixel values of the plurality of pixels in the scanned image 322 are obtained based on the input / output characteristics after the input / output characteristic adjustment processing. When a new pixel value (gradation value) is obtained based on the input / output characteristic after the input / output characteristic adjustment processing, the MFP 10 newly generates a histogram H22 (FIG. 10) with respect to the new pixel value.

FIG. 10 is a diagram showing a new histogram H22. As shown in FIG. 10, in the histogram H22 regarding the pixel values obtained based on the adjusted input / output characteristics due to the adjustment of the input / output characteristics, there are two peaks, a document region peak and an outside document region peak. Is formed more accurately.

Then, the MFP 10 calculates a temporary extraction threshold value TH22 (see FIG. 10) based on the new histogram H22, and has a value corresponding to the temporary extraction threshold value TH22 in the input / output characteristics before the input / output characteristic adjustment processing. Is calculated as a new extraction threshold TH25 for the scanned image 321.

Specifically, the input value V1 (see FIG. 9) when the temporary extraction threshold value TH22 is output is obtained based on the input / output characteristics after the input / output characteristic adjustment processing. Then, based on the input / output characteristics before the input / output characteristic adjustment processing, the output value corresponding to the input value V1 (FIG. 9) is calculated as a new extraction threshold TH25 for the scanned image 321.

When the scanned image 321 (FIG. 7) in which the difference between the pixel value of the original area and the pixel value of the outer area of the original is relatively small is generated in this way, the input / output characteristic adjustment process is executed, and the input / output characteristic is adjusted. A new extraction threshold TH25 is calculated based on a new histogram H22 of pixel values obtained based on the input / output characteristics after the adjustment process.

Here, in the input / output characteristics before the input / output characteristic adjustment processing, even if the two input values have a slight difference in A / D conversion, they are converted to the same output value (pixel value) due to the quantization error. Will be done. In other words, the input value of the specific range W1 is not converted into an appropriate output value (pixel value). As a result, for example, as shown in FIG. 8, in the histogram H21 regarding the input / output characteristics before the input / output characteristic adjustment processing, “peaks” (manuscript region peaks and extra-manuscript region peaks) and “valleys” do not appear and are accurate. The extraction threshold TH may not be calculated.

On the other hand, by performing the input / output characteristic adjustment processing, in the input / output characteristics after the input / output characteristic adjustment processing, two input values having a slight difference are converted into different output values. In other words, errors and the like associated with quantization are reduced, and quantization is performed more appropriately.

As a result, as shown in FIG. 10, in the histogram H22 regarding the input / output characteristics after the input / output characteristic adjustment processing, “mountains” and “valleys” that did not appear before the input / output characteristic adjustment processing appear. In other words, due to the input / output characteristic adjustment process, the original region peak and the extra-manuscript region peak appear in the histogram H22, and a “valley” appears between the two peaks. Therefore, it is possible to calculate an accurate extraction threshold TH25 based on the histogram H22.

When the new extraction threshold TH25 is calculated, the MFP 10 executes the main scan to generate the scanned images 400 (420) (FIG. 7). Then, the MFP 10 discriminates between the manuscript area and the non-manuscript area based on the newly calculated extraction threshold value TH25 (extraction threshold value for the entire scanned image 320). Specifically, binarization processing is executed for each pixel in the scanned image 420 based on the magnitude relationship with the new extraction threshold TH25, and edge candidates of the original region are extracted based on the processing result. .. When the edge candidates of the document area are extracted, the closed area formed by the edge candidates is determined as the document area in the scanned image 420, and the area other than the closed area is determined as the outside document area in the scanned image 420.

When the document region and the region outside the document are discriminated by the scanned image 420, the second discriminating method ends, and the process proceeds to step S20. Then, the area outside the document is erased from the scanned image 420 (step S20).

As described above, the condition that the gradation of a predetermined degree or more does not exist in the scanned image 300 and the distance between the original region peak and the extra-manuscript region peak is larger than a certain degree is not satisfied in the histogram related to the scanned image 300. In this case, the input / output characteristic adjustment process is executed for the pixel values of the plurality of pixels in the scanned image 300. Then, a new extraction threshold value TH25 is calculated based on the new histogram H22 of the pixel values obtained based on the input / output characteristics after the input / output characteristic adjustment processing. In short, the input / output characteristic adjustment process generates a more accurate histogram for the scanned image 300, and a more accurate extraction threshold TH25 is calculated based on the histogram. Therefore, even when the difference between the pixel value of the pixel in the document area and the pixel value of the pixel in the outside area of the document is smaller than a certain level in the scanned image 300, the document area and the outside area of the document can be more accurately determined. It is possible to discriminate (good).

<About the third discrimination method (area division method)>
By the way, in the first and second discrimination methods, the original region and the non-manuscript region are discriminated based on the extraction threshold value TH for the entire scanned image 300.

Here, when the document cover is in the half-opened state (a state in which the document cover is closed by about half with respect to the fully opened state), as described in the above-mentioned problem column, a gradation of a predetermined degree or more is generated in the scanned image 300, and the scanned image is scanned. It may not be possible to accurately calculate the extraction threshold TH for the entire 300. As a result, the document area and the area outside the document may not be accurately discriminated.

Specifically, as shown in FIG. 11, when the document cover 110 is in the half-open state, a part of the light emitted from the light source 150 and transmitted through the glass surface for placing the document is partly transmitted through the reflector 120 of the document cover 110. The reflected light is transmitted again through the glass surface for placing the document and is incident on the light receiving portion. In this case, a relatively large amount of reflected light is incident on one side of the platen 140 (here, the back side in the depth direction (the side closer to the hinge portion 130)), and the reflected light is almost incident on the other side (front side). do not do. Therefore, in the scanned image 300 (330) relating to the original, as shown in FIG. 12, a gradation in the vertical direction (main scanning direction) (direction orthogonal to the rotation axis of the hinge portion 130 (FIG. 11)) occurs. .. Specifically, in the scanned image 330 (see FIG. 12), the brightness value of the portion where a relatively large amount of reflected light is incident (the back side in the depth direction of the glass surface for placing the document (the side closer to the hinge portion 130)) is measured. A large gradation in which the pixel value (luminance value) gradually decreases as the position in the vertical direction (main scanning direction) moves from the back side to the front side (from the side close to the hinge portion 130 to the side close to the operation panel portion 6c). Occurs.

In the histogram H30 for the entire scanned image 330 having such a gradation, two peaks, a peak corresponding to the document region and a peak corresponding to the outside region, are not formed, but a single peak is formed (FIG. 6). See 13). In the histogram H30 having the single peak, it is difficult to specify the position (pixel value) to be set as the extraction threshold TH. Further, if the threshold value TH30 is set at a provisional position, the original area and the outer area of the original cannot be accurately discriminated. For example, when the extraction threshold TH30 is tentatively set in the left foot portion of a single peak (see FIG. 13), both pixels of the manuscript area and the non-manuscript area are mixed in the single peak. As a result, in the scanned image, as shown in FIG. 23, the original area and the outer area of the original cannot be accurately discriminated.

In consideration of this point, when the scanned image 300 has a gradation of a predetermined degree or more and the change direction of the gradation of the predetermined degree or more is the vertical direction or the horizontal direction, a third discrimination method (region division method) ) Is used to distinguish between the original area and the outer area. Specifically, as will be described later, the scanned image 300 (330) is divided in the changing direction of the gradation and divided into a plurality of divided regions 380, and the extraction threshold value TH (TH30) is calculated for each divided region 380. Will be done. Then, edge candidates of the document region are extracted in each division region 380 based on the extraction threshold TH30 for each division region 380, and the document region and the region outside the document are discriminated.

Here, the user places the document on the glass surface for placing the document, and then opens the document cover 110 in a half-opened state to give a job execution instruction.

In the MFP 10, a pre-scan is executed in response to the execution instruction (step S11), and a scan image 330 (FIG. 12) is generated.

Then, the original area and the outer area of the original are estimated in the scanned image 330 (step S12). Specifically, a histogram H30 (see FIG. 13) is generated for the pixel values of a plurality of pixels in the entire scanned image 330, and an extraction threshold TH30 for extracting edge candidates in the document region is set based on the histogram H30. Temporarily set. Here, as described above, a provisional extraction threshold TH30 is set (temporarily set) at the left foot portion of a single peak. Then, in the scanned image 330, the original region and the extra-manuscript region are estimated based on the provisionally calculated extraction threshold TH30 (see also FIG. 23).

When the document area and the area outside the document are estimated (step S12), the distribution of pixel values of the plurality of pixels in the scanned image 330 is analyzed (steps S13 to S15).

First, in step S13, it is determined whether or not there is a gradation of a predetermined degree or more in the entire scanned image 330. Specifically, it is determined whether or not there is a gradation of a predetermined degree or more in at least one of the vertical pixel row and the horizontal pixel row in the scanned image 330.

Here, with respect to the vertical pixel sequence of interest (for example, the vertical pixel string at the left end) including a specific pixel array in the vertical direction in the region outside the document estimated from the scanned image 330 (FIG. 12), the vertical pixel array of interest. It is determined that the pixel values of the plurality of pixels in the extension direction gradually change and the fluctuation range of the pixel values of the plurality of pixels exceeds a predetermined degree (meaning that the plurality of pixels have a predetermined change characteristic). In other words, it is determined that the vertical pixel sequence of the scanned image 330 has a gradation of a predetermined degree or more. Here, the attention horizontal pixel sequence (for example, the horizontal pixel sequence at the upper end) including the specific pixel sequence in the horizontal direction in the region outside the document estimated by the scanned image 330 has the predetermined change characteristic. It is determined not to do so. In other words, it is determined that there is no gradation of a predetermined degree or more in the horizontal pixel sequence of the scanned image 330.

When at least one of the vertical pixel row and the horizontal pixel row (here, the vertical pixel row) in the scanned image 330 has a gradation of a predetermined degree or more, the entire scanned image 330 has a gradation of a predetermined degree or more. It is determined in step S13 that the process is to be performed, and the process proceeds to step S15.

In step S15, the change direction of the gradation of the predetermined degree or more is determined based on the determination result of the presence / absence determination of the gradation of the predetermined degree or more in both the vertical pixel row and the horizontal pixel row in the scanned image 300 (330). It is analyzed (judged). Specifically, it is determined whether the change direction of the gradation of a predetermined degree or more in the entire scanned image 330 is the vertical direction, the horizontal direction, or the diagonal direction.

Here, as described above, although the vertical pixel array of the scanned image 330 (FIG. 12) has a gradation of a predetermined degree or more, the horizontal pixel array does not have a gradation of a predetermined degree or more. .. In this case, the MFP 10 determines that the change direction of the gradation of a predetermined degree or more in the entire scanned image 330 is the vertical direction. Then, it is determined in step S15 that the change direction of the gradation of the predetermined degree or more in the entire scanned image 330 is the vertical direction, the process proceeds to step S18, and the third determination method (region division method) is performed. It is determined (adopted) as the most suitable optimum method for the scanned image 330 (430 (described later)).

In the third discrimination method (region division method), the scanned image 300 (330) is divided into a plurality of division regions 380, and the extraction threshold value TH (TH30) is calculated for each division region 380. Then, based on the extraction threshold value TH30 for each division area 380, the original area and the non-manuscript area are discriminated in each division area 380.

Specifically, when the direction of the gradation of a predetermined degree or more in the entire scanned image 330 is the vertical direction, the MFP 10 divides the scanned image 330 into the direction of the gradation (that is, the vertical direction) and divides the scanned image 330 into a plurality of divided regions. It is divided into 380 (for example, a predetermined number of division areas 381 to 384) (see FIG. 14). Here, the scanned image 330 is divided into four based on a predetermined number of divisions (for example, four).

When the gradation change direction in the entire scanned image 300 is the horizontal direction, the scanned image 300 is divided in the horizontal direction. For example, when the hinge portion 130 for opening and closing the document cover 110 is provided at either end (for example, the left end) in the left-right direction of the platen 140 (that is, the rotation axis of the hinge 130 is relative to the main scanning direction). The scanned image 300 (335) generated by scanning the document cover 110 in a half-opened state has a lateral gradation (see FIG. 24). In this case, the scanned image 335 is divided in the direction of the gradation (here, the horizontal direction) and divided into a plurality of divided areas (see FIG. 24).

When the scanned image 330 is divided, the MFP 10 generates histograms H31 to H34 (see FIGS. 14 and 15) relating to the pixel values of each pixel in each of the divided areas 381 to 384, and for each of the divided areas 381 to 384. The extraction thresholds TH31 to TH34 are calculated.

Here, in any of the divided regions 381 to 384, the pixel value of the pixels (the pixels on the base of the original) in the original region does not (basically) change (is constant).

In consideration of this point, the MFP 10 uses a peak (also referred to as a common position peak) having its peak position (position corresponding to the peak value (pixel value)) at a common position (pixel value) in two or more histograms. Judged as a peak corresponding to the original area (original area peak). Further, the MFP 10 has the highest frequency (maximum frequency) among the peaks other than the common position peak (peaks having the peak position at a position different from the common position) (also referred to as non-common position peaks). The common position peak is determined as a peak corresponding to the region outside the document (peak outside the document).

Specifically, as shown in FIG. 15, in the histograms H31, H33, and H34, peaks exist at common positions, respectively. In this case, the MFP 10 detects a peak existing at the common position (common position peak), and determines the common position peak as a document region peak in the histograms H31, H33, and H34. Further, among the plurality of peaks, peaks other than the common position peak are detected as non-common position peaks, and the peak having the highest frequency (maximum frequency) among the non-common position peaks is determined as the out-of-manuscript region peak.

For example, in the histograms H33 and H34, the peak on the right side of the two peaks is determined as the original region peak, and the peak on the left side is determined as the extra-manuscript region peak. Further, in the histogram H31, the peak on the left side of the two peaks is determined as the peak in the document region, and the peak on the right side is determined as the peak in the region outside the document.

Then, based on each of the histograms H31, H33, and H34 (histograms having two or more peaks), the extraction thresholds TH31, TH33 for each division region 381, 383, 384 corresponding to each histogram H31, H33, H34. , TH34 is calculated. Specifically, between the common position peak and the non-common position peak in each histogram H31, H33, H34, the specific position in the foot portion of the common position peak is the extraction threshold TH31 for the divided regions 381, 383, 384. , TH33, TH34.

Further, in the histogram H32 relating to the divided region 382, there are no two or more peaks due to the small difference between the pixels in the document region and the pixels in the region outside the document (see FIG. 15). In this case, the MFP 10 refers to the extraction threshold TH32 for the histogram H32 in which two or more peaks do not exist, and the extraction threshold TH31, for the division region 380 (here, 381, 383) adjacent to the division region 382 corresponding to the histogram H32. Calculate (estimate) based on TH33. For example, the average values of the extraction thresholds TH31 and TH33 calculated for the divided regions 381 and 383 are obtained, and the average values are calculated (estimated) as the extraction thresholds TH32 for the divided regions 382.

When the extraction thresholds TH31 to TH34 are calculated for all the divided regions 380 (381 to 384), the MFP 10 executes the main scan to generate the scanned image 400 (430) (see FIG. 14).

Then, in each of the divided regions 380 (381 to 384) in the scanned image 430, the original region and the non-manuscript region are discriminated based on the extraction thresholds TH31 to TH34 for each divided region 380. Specifically, in each divided region 380 in the scanned image 430, a pixel group having a pixel value larger than the extraction threshold TH calculated in each histogram (FIG. 15) and a pixel value smaller than the extraction threshold TH. Of the two pixel groups with the pixel group having the above, one pixel group including the common position peak is determined as the pixel group in the document region.

For example, in the histogram H34 (FIG. 15) of the divided region 384, a common position peak exists on the right side of the extraction threshold TH34 calculated by the histogram H34. Therefore, in the divided region 384 in the scanned image 430, a pixel group having a pixel value larger than the extraction threshold TH34 (a pixel group including a common position peak) is determined as a pixel group in the document region. Then, in the divided region 384 (see FIG. 14) in the scanned image 430, binarization processing is executed for each pixel in the divided region 384 based on the magnitude relationship with the extraction threshold TH34. Further, the edge extraction process is executed based on the processing result of the binarization process and the like, and edge candidates of the document area in the divided area 384 are extracted.

Similarly, in the histogram H33 (FIG. 15) of the divided region 383, the common position peak exists on the right side of the extraction threshold TH33 calculated by the histogram H33. Therefore, even in the divided region 383 in the scanned image 430, a pixel group having a pixel value larger than the extraction threshold TH33 is determined as a pixel group in the document region. Then, in the divided region 383 (see FIG. 14) in the scanned image 430, binarization processing is executed for each pixel in the divided region 383 based on the magnitude relationship with the extraction threshold TH33, and the binarization is performed. Edge candidates of the original area in the divided area 383 are extracted based on the processing result of the processing and the like.

On the other hand, as shown in FIG. 15, in the histogram H31 of the divided region 381, the common position peak exists on the left side of the extraction threshold TH31 calculated by the histogram H31. Therefore, in the divided region 381 in the scanned image 430, a pixel group having a pixel value smaller than the extraction threshold TH31 is determined as a pixel group in the document region. Then, in the divided region 381 (see FIG. 14) in the scanned image 430, a binarization process is executed for each pixel in the divided region 381 based on the magnitude relationship with the extraction threshold TH31, and the binarization process is performed. Edge candidates of the document area in the division area 381 are extracted based on the processing result and the like.

Regarding the divided region 382, the position of the common position peak specified in each histogram of the other divided regions 380 exists on the right side of the extraction threshold TH32 calculated (estimated) for the divided region 382 (). (See FIG. 15). Therefore, in the divided region 382 in the scanned image 430, a pixel group having a pixel value of a pixel larger than the extraction threshold TH32 is determined as a pixel group in the document region. Then, in the divided region 382 (see FIG. 14) in the scanned image 430, a binarization process is executed for the pixels in the divided region 382 based on the magnitude relationship with the extraction threshold TH32, and the binarization process is performed. Edge candidates of the document area in the division area 382 are extracted based on the processing result and the like.

In this way, edge candidates of the document region are extracted in each divided region 380 in the scanned image 430. Then, based on the edge candidates extracted for each of the divided regions 380, the original region and the outer region of the original are discriminated in the scanned image 430. Specifically, the closed region formed by the edge candidates extracted for each division region 380 is determined as the original region, and the region outside the closed region is determined as the outer region.

When the document region and the region outside the document are discriminated by the scanned image 430, the third discriminating method ends, and the process proceeds to step S20. Then, the area outside the document is erased from the scanned image 430 (step S20).

As described above, when the entire scanned image 300 has a gradation of a predetermined degree or more and the direction of the gradation of the predetermined degree or more is either the vertical direction or the horizontal direction, the scanned image 300 changes the gradation direction. It is divided into a plurality of division areas 380. Then, the extraction threshold value TH30 is calculated for each division area 380, and the original area and the non-manuscript area are discriminated based on the extraction threshold value TH30 for each division area 380. Therefore, even when the scanned image 300 has a gradation of a predetermined degree or more (specifically, a gradation in the vertical direction or the horizontal direction), the original area and the outer area of the original can be discriminated more accurately (goodly). Is possible.

<About the fourth discrimination method (orthogonal point estimation method)>
Here, even when the document cover 110 is fully opened, a gradation of a predetermined degree or more may occur in the entire scanned image 300.

For example, when the scanning operation is performed while the user is looking into the glass surface for placing a document, a part of the light emitted from the light source 150 and transmitted through the glass surface for placing the document is part of the user's clothing or the like. May hit. In this case, depending on the position of the user, a relatively large amount of reflected light (light that hits the user's clothing or the like) passes through the glass surface for placing the document again, and both ends on the front side of the platen 140. It may be incident on one of the corners and detected by the light receiving part. For example, when the user is looking into the glass surface for placing a document from the left front side of the platen 140 (the front side in the depth direction and the left side in the left-right direction), there are relatively many in the corner portion on the left front side of the platen 140. (Light that hits the user's clothing, etc.) is detected. In such a situation, the scanned image 300 (340) generated with respect to the document may have an oblique gradation as shown in FIG. Specifically, in the scanned image 340 (FIG. 16), the brightness value of the left front side portion of the document mounting glass surface is large, and the pixel positions are from the left front side to the right back side (depth of the document mounting glass surface). A gradation (diagonal gradation) in which the brightness value gradually decreases (darkens) may occur as the brightness value gradually decreases (darkens) toward the back side in the direction and the right side in the left-right direction.

In the scanned image 340 (FIG. 16) having such a gradation, if the scanned image 340 is divided in the vertical direction or the horizontal direction, the original area and the outer area of the original cannot be accurately discriminated in a relatively large number of divided areas. Sometimes.

Here, as described above, when the change direction of the gradation of a predetermined degree or more in the entire scanned image 300 is the vertical direction or the horizontal direction, two peaks are not detected in the histogram of the entire scanned image 300, and the histogram is the same. There may be a problem that an accurate extraction threshold TH is not calculated based on the above.

When the change direction of the gradation of a predetermined degree or more in the entire scanned image 300 is an oblique direction, even if the scanned image 300 is divided in the vertical direction, for example, in a relatively large number of divided regions, a predetermined degree in the horizontal direction. The above gradation can still exist. For example, even if the scanned image 340 having a gradation of a predetermined degree or more in the diagonal direction is vertically divided and divided into four divided areas 391-394, for example, in the three divided areas 392-394, FIG. 26 As shown above, there is a gradation of a predetermined degree or more in the horizontal direction.

Therefore, due to the presence of a gradation of a predetermined degree or more in the lateral direction in the divided regions 392 to 394, the same problem as the above problem may occur in each of the divided regions 392 to 394. For example, in each of the histograms for the divided regions 392-394, two peaks are not detected and an accurate extraction threshold TH is not calculated. As a result, in each of the divided regions 392 to 394, the document region and the region outside the document may not be accurately discriminated.

In consideration of this point, when there is a gradation of a predetermined degree or more in the entire scanned image 300 and the change direction of the gradation of the predetermined degree or more is an oblique direction, the third discrimination method (region division method) is used. Uses a different fourth discrimination method (orthogonal point estimation method) (described later) to discriminate between the manuscript area and the non-manuscript area.

Specifically, after the scan image 340 (FIG. 16) is generated in the pre-scan (step S11), the original area and the non-manuscript area are estimated in the scanned image 340 (step S12). FIG. 17 is a diagram showing a histogram H40 generated with respect to the pixel value of each pixel of the entire scanned image 340. Then, the extraction threshold TH (TH40) is tentatively set based on the histogram H40. Then, the manuscript area and the out-of-manuscript area are estimated based on the tentatively set extraction threshold value TH40.

When the document area and the area outside the document are estimated (step S12), the distribution of the pixel values of each pixel in the scanned image 340 is analyzed (steps S13 to S15).

First, in step S13, it is determined whether or not there is a gradation of a predetermined degree or more in the entire scanned image 340. Specifically, it is determined whether or not there is a gradation of a predetermined degree or more in at least one of the vertical pixel row and the horizontal pixel row in the scanned image 330.

Here, with respect to the vertical pixel sequence of interest (for example, the vertical pixel array at the left end) in the scanned image 340 (see FIG. 16), the pixel values of the plurality of pixels in the extension direction of the vertical pixel array of interest gradually change. It is determined that the fluctuation range of the pixel values of the plurality of pixels exceeds a predetermined degree (meaning that the plurality of pixels have a predetermined change characteristic). In other words, it is determined that the attention vertical pixel sequence of the scanned image 340 has a gradation of a predetermined degree or more. Further, it is determined that the attention lateral pixel array (for example, the lateral pixel array at the lower end) in the scanned image 340 also has the predetermined change characteristic. In other words, it is determined that there is a gradation of a predetermined degree or more even in the attention horizontal pixel sequence of the scanned image 340.

Then, when at least one (here, both) of the vertical pixel row and the horizontal pixel row in the scanned image 340 has a gradation of a predetermined degree or more, it means that the entire scanned image 340 has a gradation of a predetermined degree or more. The determination is made in step S13, and the process proceeds to step S15.

In step S15, the change direction of the gradation of the predetermined degree or more is determined based on the determination result of the presence / absence determination of the gradation of the predetermined degree or more in both the vertical pixel row and the horizontal pixel row in the scanned image 340. .. Specifically, it is determined whether the change direction of the gradation of a predetermined degree or more in the entire scanned image 340 is the vertical direction, the horizontal direction, or the diagonal direction.

Here, as described above, a gradation of a predetermined degree or more exists in both the vertical pixel array and the horizontal pixel array of the scanned image 340 (FIG. 16). In this case, the MFP 10 determines that the change direction of the gradation of a predetermined degree or more in the entire scanned image 340 is an oblique direction. Then, it is determined in step S15 that the change direction is an oblique direction, the process proceeds to step S19, and a fourth discrimination method (orthogonal point estimation method) different from the third discrimination method is used for the scanned image. It is determined (adopted) as the most suitable optimum method for 340 (440 (described later)).

In the fourth discrimination method (orthogonal point estimation method), when the document region and the outer region of the document are discriminated based on the positions of the four orthogonal points P1 to P4 where the edge candidates of the document region are orthogonal to each other, the four points are used. When it is determined that at least one of the orthogonal points P1 to P4 is not detected, the position of the one orthogonal point is determined based on the detection position of the remaining orthogonal points. Then, the original region and the outer region of the original are discriminated based on the closed region formed by the straight line connecting the orthogonal points P1 to P4.

Specifically, first, the MFP 10 executes the main scan to generate a scanned image 400 (440) (see FIG. 18). Then, the MFP 10 uses the extraction threshold value TH40 (extraction threshold value TH40 calculated for the entire scanned image 340) tentatively calculated (temporarily set) in step S12 as it is, and uses the edge of the document region in the scanned image 440 as it is. Extract candidates. When the edge candidates are extracted, the MFP 10 executes an orthogonal point detection process for detecting the orthogonal points P to which the extracted edge candidates are orthogonal to each other.

Here, when a diagonal gradation is present in the scanned image 400, three of the four orthogonal points of the edge candidates are relatively likely to be detected, but among the four orthogonal points. It is relatively likely that one point will not be detected. Specifically, the orthogonal points (here, orthogonal points P1 to P3) are detected in the region outside the document of the scanned image 440 (see FIG. 18) where the reflected light is hardly detected. On the other hand, of the document area of the scanned image 440, the portion where a relatively large amount of reflected light (light reflected by the user's clothing or the like) is incident (here, the left corner on the front side in the depth direction of the glass surface for placing the document). ) Is not detected at the orthogonal point (here, the orthogonal point P4).

In consideration of this point, it is determined that at least one orthogonal point (here, P4) among the four orthogonal points P1 to P4 is not detected when the scanned image 440 has a gradient in the oblique direction. Occasionally, the position of the orthogonal point P4 is determined (estimated) based on the detection position of the remaining orthogonal points P1 to P3.

Specifically, the closed region formed by connecting the orthogonal points P1 to P4 with a straight line has a rectangular shape based on the positional relationship and the geometrical relationship of the detection positions of the orthogonal points P1 to P3. , The position of the orthogonal point P4 is estimated (see FIG. 19).

Then, in the scanned image 440, a closed region is formed based on the detection positions of the orthogonal points P1 to P3 and the estimated position of the orthogonal points P4, the closed region is determined as the original region, and the region other than the closed region is determined. Is determined as the area outside the document.

When the document region and the region outside the document are discriminated by the scanned image 440, the fourth discrimination method (orthogonal point estimation method) ends, and the process proceeds to step S20. Then, the area outside the document is erased from the scanned image 440 (step S20).

As described above, when the entire scanned image 300 has a gradation of a predetermined degree or more and the change direction of the gradation of the predetermined degree or more is an oblique direction, the document is based on the position of the orthogonal point of the edge candidate in the document area. The area and the area outside the document are discriminated. However, when such a determination is made, if the undetected orthogonal point (here, the orthogonal point P4) exists, the position of the orthogonal point P4 is the remaining orthogonal point (here, the orthogonal points P1 to P3). Estimated based on the detection position. Therefore, even when there is a gradation in the oblique direction in the scanned image 300, it is possible to more accurately (satisfy) discriminate between the document region and the region outside the document.

As described above, in the present embodiment, the distribution of the pixel values of the plurality of pixels in the scanned image of the document placed on the platen 140 is analyzed (steps S13 to S15), and based on the analysis result, the distribution state is analyzed. The optimum method is determined from a plurality of discrimination methods (steps S16 to S19) for discriminating between the document area and the area outside the document. Then, the manuscript area and the non-manuscript area are discriminated by using the determined optimum method. Therefore, in the scanned image of the document placed on the platen, the document area and the area outside the document can be more accurately discriminated.

<1-3. Deformation example, etc.>
Although the embodiments of the present invention have been described above, the present invention is not limited to the above contents.

<Examples of modifications related to the pros and cons of executing each discrimination method>
For example, in the above embodiment, after the presence / absence determination process (step S13) of the gradation of a predetermined degree or more in the scanned image 300 is performed, the frequency distribution state analysis process (step S14) in the histogram of the entire scanned image 300 is performed. However, it is not limited to this. For example, the presence / absence determination process of the gradation may be performed after the analysis process of the frequency distribution situation is performed.

Further, in the above embodiment and the like, all three processes of the presence / absence determination process of the gradation (step S13), the analysis process of the frequency distribution status (step S14), and the direction determination process of the gradation (step S15) are performed. However, the present invention is not limited to this, and only a part of the three processes may be performed.

For example, the frequency distribution status analysis process (step S14) may not be performed, but the gradation presence / absence determination process (step S13) and the gradation direction determination process (step S15) may be performed.

Specifically, when the scanned image 300 does not have a gradation of a predetermined degree or more, the frequency distribution status analysis process is not performed, and the first discrimination method (basic discrimination method) and the second discrimination method (input / output characteristic adjustment) are performed. The discrimination method of either one of the method) may be uniformly determined (adopted) as the optimum method. If the entire scanned image 300 has a gradation of a predetermined degree or more, the gradation direction determination process is executed in the same manner as in the above embodiment, and a third determination method (region division method) is performed according to the determination result. And one of the fourth discrimination method (orthogonal point estimation method) is determined as the optimum method.

Alternatively, the presence / absence determination process of the gradation (step S13) and the direction determination process of the gradation (step S15) may not be performed, but only the analysis process of the frequency distribution status (step S14) may be performed. Specifically, the presence or absence of a gradation of a predetermined degree or more in the scanned image 300 is not determined, and the condition that the distance between the original region peak and the outer region peak in the original region is larger than a certain degree in the histogram of the entire scanned image 300. May only be determined whether or not is satisfied.

Further, the gradation direction determination process (step S15) may not be performed, but the gradation existence / absence determination process (step S13) and the frequency distribution status analysis process (step S14) may be performed. Specifically, when it is determined that the scanned image 300 has a gradation of a predetermined degree or more, the direction determination process of the gradation is not performed, and the third discrimination method (region division method) and the fourth discrimination method Either one of (orthogonal point estimation method) may be uniformly determined (adopted) as the optimum method.

In this way, even if only a part of the three processes of the gradation existence / absence determination process (step S13), the frequency distribution status analysis process (step S14), and the gradation direction determination process (step S15) is performed. Good.

<Modification example of the second discrimination method (step S17)>
Further, in the above-described embodiment or the like, when the condition that the distance between the document region peak and the document outside region peak is larger than a certain level is not satisfied in the histogram, the input / output characteristic adjustment method is the second discrimination method (step S17). ), But is not limited to this. For example, when the condition that the distance between the original region peak and the outer region peak is larger than a certain degree is not satisfied in the histogram, the following discrimination method may be used as the second discrimination method.

Specifically, the complement processing based on the regular edge portion (described later) of the extracted edge candidates is performed to correct the false detection portion (described later) of the edge candidate, and the manuscript area and the non-manuscript area are set. The discriminating method (“edge complementing method”) may be used as the second discriminating method (step S17).

Specifically, the MFP 10 extracts edge candidates of the manuscript area based on the extraction threshold value TH tentatively calculated in step S12, and mistakenly mistakes the extracted edge candidates as the normal edge portion of the manuscript area. Identify the detection part. Then, the MFP 10 performs complementary processing based on the regular edge portion to correct the false detection portion, and discriminates between the original region and the outer region of the original.

More specifically, in step S17, the MFP 10 executes the main scan to generate a scanned image 420 (see FIG. 7). Then, the MFP 10 uses the extraction threshold value TH21 (extraction threshold value calculated based on the histogram H21 of the pixel values of the plurality of pixels in the scanned image 320) (FIG. 8) tentatively calculated in step S12 as it is. Then, the edge candidates of the document area are extracted from the scanned image 425.

However, if the condition that the distance between the original region peak and the outer region peak is larger than a certain level is not satisfied, the extraction threshold TH is not calculated accurately, and a part of the edge candidates is shown, for example, as shown in FIG. May not be extracted accurately. In FIG. 20, for convenience of illustration, both the original area and the outer area of the original are displayed in white, but in reality, as shown in FIG. 7, the pixel values of the pixels in the outer area of the original are relatively small (as shown in FIG. 7). (Almost close to the value "0"), it is assumed that the pixel value of the pixel in the document region is slightly larger than the pixel value of the pixel in the region outside the document.

In consideration of this point, the MFP 10 is based on the regular edge portion of the extracted edge candidates when the condition that the distance between the original region peak and the outer region peak is larger than a certain degree is not satisfied. A method (“edge complementation method”) for discriminating between the original area and the non-manuscript area by performing the complement processing to correct the erroneously detected portion of the edge candidate is determined (adopted) as the optimum method.

Specifically, the MFP 10 sets a straight line portion (line segment) having a length equal to or longer than a predetermined value among the extracted edge candidates as a regular edge portion (a portion deemed to be correctly detected) in the document area. Identify (see FIG. 21). Here, the straight line portions M1 to M7 (see FIG. 21) are specified as regular edge portions. Further, the MFP 10 extracts a portion (a portion having no linearity or a linear line segment having a length less than the predetermined value) other than the linear portion (regular edge portion) having a length equal to or longer than the predetermined value as an erroneous detection portion. .. Here, four false detection portions N1 to N4 (see FIG. 21) are specified.

Then, the MFP 10 performs complementary processing based on the straight line portions M1 to M7 specified as the regular edge portions to correct the false detection portions N1 to N4 (see FIG. 21). For example, the straight line portion M1 (and / or M2) is extended to correct the false positive portion N1. Further, the straight line portion M3 (and / or M4) is extended to correct the false detection portion N2. Further, the straight line portion M5 and the straight line portion M6 are extended to correct the erroneous detection portion N3. Further, the linear portion M6 (and / or M7) is extended to correct the false positive portion N4.

When all the false detection portions N1 to N4 are corrected, the MFP 10 determines in the scanned image 420 that the closed region formed by the corrected edge candidates is the original region, and the region other than the closed region is the outer region of the original. To determine as.

In this way, when the condition that the distance between the original region peak and the outer region peak is larger than a certain level is not satisfied in the histogram, the false detection portion is corrected by the complement processing based on the regular edge portion of the original region. The method (edge complementation method) may be used as the second discrimination method (step S17).

<Modification example of the fourth discrimination method (step S19)>
Further, in the above-described embodiment or the like, when the change direction of the gradation of a predetermined degree or more in the entire scanned image 300 is the oblique direction in the scanned image 300, the orthogonal point estimation method is used as the fourth determination method (step S19). However, it is not limited to this. For example, when the change direction of the gradation of a predetermined degree or more in the entire scanned image 300 is an oblique direction, the above edge complementing method may be used as the fourth determination method.

Specifically, in step S19, the MFP 10 executes the main scan to generate a scanned image 440 (see FIG. 16). Then, the MFP 10 uses the extraction threshold value TH40 (extraction threshold value calculated based on the histogram H40 of the pixel values of the plurality of pixels in the scanned image 340) tentatively calculated in step S12 (FIG. 17) as it is. Then, edge candidates of the document region are extracted from the scanned image 440 (see FIG. 22).

When the edge candidates are extracted, the MFP 10 identifies a normal edge portion (a straight line portion having a length equal to or longer than a predetermined value) among the extracted edge candidates, and a false detection portion (a portion other than the regular edge portion). (Here, the false detection portion N10) is specified.

Then, the MFP 10 corrects the erroneous detection portion N10 by performing complementary processing based on the regular edge portion. Specifically, the straight line portion M11 and the straight line portion M12 are extended to correct the false detection portion N10 (see FIG. 22). When the erroneous detection portion N10 is corrected, the MFP 10 determines in the scanned image 440 the closed region formed by the corrected edge candidate as the original region, and determines the region other than the closed region as the outer region of the original. To do.

As described above, when the change direction of the gradation in the entire scanned image 300 is an oblique direction, the edge complementing method may be used as the fourth determination method (step S19).

<Others>
Further, in the above-described embodiment and the like, the first to fourth discrimination methods (steps S16 to S19) are different discrimination methods from each other, but the present invention is not limited to this, and one of the first to fourth discrimination methods. The parts may be the same discrimination method. For example, the second discrimination method and the fourth discrimination method may be the same discrimination method. Specifically, in step S17 and step S19, the same discrimination method (the above-mentioned edge complementation method, orthogonal point estimation process, etc.) may be determined as the optimum method.

Further, in each of the above embodiments, the scan image 300 generated by the pre-scan is used as an image for analyzing the distribution of pixel values of a plurality of pixels in the scan image 300, and is generated by the main scan. The scanned image 400 is used as a processing target image for the non-manuscript erasing process, but is not limited thereto. For example, the scanned image 300 generated by the pre-scan may be used not only as an image for analysis of the distribution state but also as a processing target image of the non-manuscript erasing process.

10 Image forming apparatus (MFP)
110 Original cover 140 Original stand 300, 310, 320, 330, 340 Scanned image (during pre-scan)
380 to 384 divided areas 400, 410, 420, 430, 440 Scanned image (at the time of main scan)

Claims (34)

It is an image processing device
An analytical means for analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen, and an analysis means.
Based on the analysis result by the analysis means, from among a plurality of discrimination methods for discriminating between the manuscript area which is the region where the manuscript exists and the region outside the manuscript which is the region where the manuscript does not exist in the image. An area determination means for determining an optimum method, which is the most suitable method for the scanned image, and using the determined optimum method to discriminate between the original area and the outer area.
With
The plurality of discrimination methods include a first discrimination method and a third discrimination method.
In the first discrimination method, among a plurality of peaks existing in the frequency distribution relating to the pixel values of a plurality of pixels in the entire scanned image, the peak having the smallest pixel value at the peak position corresponds to the out-of-manuscript region. It is determined as the peak outside the original region, and the residual peak is determined as the peak in the original region corresponding to the original region, and the pixel value of the peak outside the original and the pixel value of the peak in the original region. It is a method of discriminating between the manuscript area and the out-of-manuscript area based on the threshold value.
In the third discrimination method, the scanned image is divided into one of the vertical direction and the horizontal direction and divided into a plurality of divided regions, and the frequency distribution for each divided region is within the divided region. Of the plurality of peaks existing in the frequency distribution relating to the pixel values of the plurality of pixels, the value between the pixel value of the out-of-manuscript region peak and the pixel value of the manuscript region peak is set as the threshold value for each division region, and for each division region. It is a method of discriminating between the original area and the outer area of the original for each divided area based on the threshold value.
The analysis means determines whether or not there is a gradation of a predetermined degree or more in the entire scanned image based on the spatial distribution of the pixel values of the plurality of pixels in the scanned image.
The area determination means
The third discrimination method is determined as the optimum method on condition that a gradation of a predetermined degree or more is present in the entire scanned image.
An image processing apparatus characterized in that the first discrimination method is determined as the optimum method on condition that a gradation of a predetermined degree or more does not exist in the entire scanned image. In the image processing apparatus according to claim 1,
The analysis means
When the scanned image has a gradation of the predetermined degree or more, the change direction of the gradation of the predetermined degree or more is further determined based on the spatial distribution situation.
The area determination means
The third determination method is performed on condition that it is determined that there is a gradation of a predetermined degree or more in the entire scanned image and that the changing direction is either the vertical direction or the horizontal direction in the scanned image. An image processing apparatus characterized in that it is determined as the optimum method and the change direction of the gradation of a predetermined degree or more is determined as the one direction in the third discrimination method. In the image processing apparatus according to claim 2,
The plurality of discrimination methods also include a fourth discrimination method.
The fourth discrimination method is
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates in the document region, and the document is based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the region from the extra-manuscript region, the positional relationship and geometry of the detection positions of the remaining orthogonal points. Based on the relationship, the positions of at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is used as the original document. A method of discriminating as an area and determining an area other than the closed area as the extra document area,
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold value to extract edge candidates in the original area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value. Is specified as a regular edge portion of the original document area, a portion other than the straight line portion is specified as a false detection portion, and complementary processing based on the regular edge portion is performed to correct the false detection portion and correct it. It is one of the methods of discriminating the closed region formed by the later edge candidate as the document region and determining the region other than the closed region as the outer region of the document.
The area discrimination means uses the fourth discrimination method on the condition that a gradation of a predetermined degree or more is present in the entire scanned image and it is determined that the change direction is an oblique direction in the scan image. An image processing device characterized in that it is determined as an optimum method. It is an image processing device
An analytical means for analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen, and an analysis means.
Based on the analysis result by the analysis means, from among a plurality of discrimination methods for discriminating between the manuscript area which is the region where the manuscript exists and the region outside the manuscript which is the region where the manuscript does not exist in the image. An area determination means for determining an optimum method, which is the most suitable method for the scanned image, and using the determined optimum method to discriminate between the original area and the outer area.
With
The plurality of discrimination methods include a first discrimination method and a third discrimination method.
In the first discrimination method, among a plurality of peaks existing in the frequency distribution relating to the pixel values of a plurality of pixels in the entire scanned image, the peak having the smallest pixel value at the peak position corresponds to the out-of-manuscript region. It is estimated as the peak of the extra-manuscript region peak, and the residual peak is estimated as the peak of the original region, which is the peak corresponding to the manuscript region, and the pixel value of the peak of the extra-manuscript region and the pixel value of the peak of the manuscript region. It is a method of discriminating between the manuscript area and the out-of-manuscript area based on the threshold value.
In the third discrimination method, the scanned image is divided into one of the vertical direction and the horizontal direction and divided into a plurality of divided regions, and the frequency distribution for each divided region is within the divided region. Of the plurality of peaks existing in the frequency distribution relating to the pixel values of the plurality of pixels, the value between the pixel value of the out-of-manuscript region peak and the pixel value of the manuscript region peak is set as the threshold value for each division region, and for each division region. It is a method of discriminating between the original area and the outer area of the original for each divided area based on the threshold value.
The analysis means determines whether or not the distance between the extra-manuscript region peak and the manuscript region peak is larger than a certain degree in the frequency distribution relating to the pixel values of the plurality of pixels in the entire scanned image.
The region discriminating means uses the first discriminating method as the optimum method on condition that it is determined in the frequency distribution that the distance between the out-of-manuscript region peak and the manuscript region peak is larger than a certain level. Determined as
The analysis means determines whether or not there is a gradation of a predetermined degree or more in the entire scanned image based on the spatial distribution of the pixel values of the plurality of pixels in the scanned image.
The image processing apparatus is characterized in that the area discrimination means determines the third discrimination method as the optimum method on condition that a gradation of a predetermined degree or more is present in the entire scanned image. In the image processing apparatus according to claim 4,
The plurality of discrimination methods also include a second discrimination method.
The second discrimination method is
This is an input / output characteristic adjustment process that adjusts the input / output characteristics between the analog input value that indicates the amount of light received by the light receiving element in the scan process and the digital output value that indicates the output pixel value, and is the input before the input / output characteristic adjustment process. Pixel values of a plurality of pixels in the entire second scan image obtained after the input / output characteristic adjustment processing for adjusting the input / output characteristics so that a part of the output range in the output characteristics becomes a wider output range. In the frequency distribution with respect to, the value between the newly obtained pixel value of the extra-manuscript region peak and the newly obtained pixel value of the manuscript region peak is set as a threshold, and the manuscript region and the outside of the manuscript are based on the threshold. The method of distinguishing from the area and
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates in the document region, and the document is based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the region from the extra-manuscript region, the positional relationship and geometry of the detection positions of the remaining orthogonal points. Based on the relationship, the positions of at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is used as the original document. A method of discriminating as an area and determining an area other than the closed area as the extra document area,
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold value to extract edge candidates in the original area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value. Is specified as a regular edge portion of the original document area, a portion other than the straight line portion is specified as a false detection portion, and complementary processing based on the regular edge portion is performed to correct the false detection portion and correct it. It is one of the methods of discriminating the closed region formed by the later edge candidate as the document region and determining the region other than the closed region as the outer region of the document.
The area discrimination means is characterized in that the second discrimination method is determined as the optimum method on condition that it is determined that the distance between the document outer region peak and the document region peak is smaller than a certain degree. Image processing device. In the image processing apparatus according to claim 4,
The analysis means
When the scanned image has a gradation of the predetermined degree or more, the change direction of the gradation of the predetermined degree or more is further determined based on the spatial distribution situation.
The area determination means
The third determination method is performed on condition that it is determined that there is a gradation of a predetermined degree or more in the entire scanned image and that the changing direction is either the vertical direction or the horizontal direction in the scanned image. An image processing apparatus characterized in that it is determined as the optimum method and the change direction of the gradation of a predetermined degree or more is determined as the one direction in the third discrimination method. In the image processing apparatus according to claim 6,
The plurality of discrimination methods also include a fourth discrimination method.
The fourth discrimination method is
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates in the document region, and the document is based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the region from the extra-manuscript region, the positional relationship and geometry of the detection positions of the remaining orthogonal points. Based on the relationship, the positions of at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is used as the original document. A method of discriminating as an area and determining an area other than the closed area as the extra document area,
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold value to extract edge candidates in the original area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value. Is specified as a regular edge portion of the original document area, a portion other than the straight line portion is specified as a false detection portion, and complementary processing based on the regular edge portion is performed to correct the false detection portion and correct it. It is one of the methods of discriminating the closed region formed by the later edge candidate as the document region and determining the region other than the closed region as the outer region of the document.
The area discrimination means uses the fourth discrimination method on the condition that a gradation of a predetermined degree or more is present in the entire scanned image and it is determined that the change direction is an oblique direction in the scan image. An image processing device characterized in that it is determined as an optimum method. In the image processing apparatus according to claim 7,
The plurality of discrimination methods also include a second discrimination method.
The second discrimination method is
This is an input / output characteristic adjustment process that adjusts the input / output characteristics between the analog input value that indicates the amount of light received by the light receiving element in the scan process and the digital output value that indicates the output pixel value, and is the input before the input / output characteristic adjustment process. Pixel values of a plurality of pixels in the entire second scan image obtained after the input / output characteristic adjustment processing for adjusting the input / output characteristics so that a part of the output range in the output characteristics becomes a wider output range. In the frequency distribution with respect to, the value between the newly obtained pixel value of the extra-manuscript region peak and the newly obtained pixel value of the manuscript region peak is set as a threshold, and the manuscript region and the outside of the manuscript are based on the threshold. The method of distinguishing from the area and
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates in the document region, and the document is based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the region from the extra-manuscript region, the positional relationship and geometry of the detection positions of the remaining orthogonal points. Based on the relationship, the positions of at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is used as the original document. A method of discriminating as an area and determining an area other than the closed area as the extra document area,
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold value to extract edge candidates in the original area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value. Is specified as a regular edge portion of the original document area, a portion other than the straight line portion is specified as a false detection portion, and complementary processing based on the regular edge portion is performed to correct the false detection portion and correct it. It is one of the methods of discriminating the closed region formed by the later edge candidate as the document region and determining the region other than the closed region as the outer region of the document.
The area discrimination means is characterized in that the second discrimination method is determined as the optimum method on condition that it is determined that the distance between the document outer region peak and the document region peak is smaller than a certain degree. Image processing device. In the image processing apparatus according to claim 8,
The image processing apparatus, wherein the second discrimination method is the same method as the fourth discrimination method. In the image processing apparatus according to claim 8,
An image processing apparatus characterized in that the second discrimination method is a method different from the fourth discrimination method. It is an image processing device
An analytical means for analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen, and an analysis means.
Based on the analysis result by the analysis means, from among a plurality of discrimination methods for discriminating between the manuscript area, which is the region where the manuscript exists in the image, and the outer region, which is the region where the manuscript does not exist, An area determination means for determining an optimum method, which is the most suitable method for the scanned image, and using the determined optimum method to discriminate between the original area and the outer area.
With
The plurality of discrimination methods include a third discrimination method and a fourth discrimination method.
In the third discrimination method, the scanned image is divided into one of the vertical direction and the horizontal direction and divided into a plurality of divided regions, and the frequency distribution for each divided region is within the divided region. Of the plurality of peaks existing in the frequency distribution relating to the pixel values of a plurality of pixels, the pixel value of the extra-manuscript region peak, which is the peak corresponding to the extra-manuscript region, and the pixel value of the manuscript region peak, which is the peak corresponding to the manuscript region It is a method of discriminating between the original area and the outer area of the original for each divided area based on the threshold value for each divided area as the value between and.
The fourth discrimination method is
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates in the document region, and the document is based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the region from the extra-manuscript region, the positional relationship and geometry of the detection positions of the remaining orthogonal points. Based on the relationship, the positions of at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is used as the original document. A method of discriminating as an area and determining an area other than the closed area as the extra document area,
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold value to extract edge candidates in the original area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value. Is specified as a regular edge portion of the original document area, a portion other than the straight line portion is specified as a false detection portion, and complementary processing based on the regular edge portion is performed to correct the false detection portion and correct it. It is one of the methods of discriminating the closed region formed by the later edge candidate as the document region and determining the region other than the closed region as the outer region of the document.
The analysis means
Based on the spatial distribution of the pixel values of the plurality of pixels in the scanned image, it is determined whether or not there is a gradation of a predetermined degree or more in the entire scanned image.
When the scanned image has a gradation of the predetermined degree or more, the change direction of the gradation of the predetermined degree or more is further determined based on the spatial distribution situation.
The area determination means
The third determination method is performed on condition that it is determined that there is a gradation of a predetermined degree or more in the entire scanned image and that the changing direction is either the vertical direction or the horizontal direction in the scanned image. It is determined as the optimum method, and the change direction of the gradation of a predetermined degree or more is determined as the one direction in the third discrimination method.
The fourth determination method is determined as the optimum method on condition that a gradation of a predetermined degree or more is present in the entire scanned image and it is determined that the change direction is an oblique direction in the scanned image. An image processing device characterized by. In the image processing apparatus according to any one of claims 1 to 11.
In the third discrimination method, the scanned image is divided into the one direction of the vertical direction and the horizontal direction and divided into the plurality of divided regions, and two or more of the plurality of divided regions are divided. Of the plurality of peaks existing in each frequency distribution, the two or more divisions of the two or more division regions having a frequency distribution of two or more peaks with respect to the pixel values of the plurality of pixels in each division region. A common position peak, which is a peak having its peak position at a common position in a frequency distribution of two or more with respect to a region, is determined as the document region peak, and a specific non-common position peak among non-common position peaks other than the common position peak is determined. The common position peak is determined as the extra-manuscript region peak, and the value between the pixel value of the extra-manuscript region peak and the pixel value of the manuscript region peak is set as the threshold value for each division region, and the threshold value for each division region is set. An image processing apparatus characterized in that it is a method of discriminating between the document area and the area outside the document for each divided area based on the method. In the image processing apparatus according to claim 12,
In the third discrimination method, the threshold value of the divided region in which the frequency distribution regarding the pixel values of the plurality of pixels does not have a peak of 2 or more is obtained based on the threshold value of the other divided regions. An image processing device characterized by the fact that. It is an image processing device
An analytical means for analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen, and an analysis means.
Based on the analysis result by the analysis means, from among a plurality of discrimination methods for discriminating between the manuscript area which is the region where the manuscript exists and the region outside the manuscript which is the region where the manuscript does not exist in the image. An area determination means for determining an optimum method, which is the most suitable method for the scanned image, and using the determined optimum method to discriminate between the original area and the outer area.
With
The plurality of discrimination methods include a first discrimination method and a second discrimination method.
In the first discrimination method, among a plurality of peaks existing in the frequency distribution relating to the pixel values of a plurality of pixels in the entire scanned image, the peak having the smallest pixel value at the peak position corresponds to the out-of-manuscript region. It is estimated as the peak of the extra-manuscript region peak, and the residual peak is estimated as the peak of the original region, which is the peak corresponding to the manuscript region, and the pixel value of the peak of the extra-manuscript region and the pixel value of the peak of the manuscript region. It is a method of discriminating between the manuscript area and the out-of-manuscript area based on the threshold value.
The second discrimination method is
This is an input / output characteristic adjustment process that adjusts the input / output characteristics between the analog input value that indicates the amount of light received by the light receiving element in the scan process and the digital output value that indicates the output pixel value, and is the input before the input / output characteristic adjustment process. Pixel values of a plurality of pixels in the entire second scan image obtained after the input / output characteristic adjustment processing for adjusting the input / output characteristics so that a part of the output range in the output characteristics becomes a wider output range. In the frequency distribution with respect to, the value between the newly obtained pixel value of the extra-manuscript region peak and the newly obtained pixel value of the manuscript region peak is set as a threshold, and the manuscript region and the outside of the manuscript are based on the threshold. The method of distinguishing from the area and
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates in the document region, and the document is based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the region from the extra-manuscript region, the positional relationship and geometry of the detection positions of the remaining orthogonal points. Based on the relationship, the positions of at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is used as the original document. A method of discriminating as an area and determining an area other than the closed area as the extra document area,
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold value to extract edge candidates in the original area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value. Is specified as a regular edge portion of the original document area, a portion other than the straight line portion is specified as a false detection portion, and complementary processing based on the regular edge portion is performed to correct the false detection portion and correct it. It is one of the methods of discriminating the closed region formed by the later edge candidate as the document region and determining the region other than the closed region as the outer region of the document.
The analysis means determines whether or not the distance between the extra-manuscript region peak and the manuscript region peak is larger than a certain degree in the frequency distribution relating to the pixel values of the plurality of pixels in the entire scanned image .
The area determination means has a frequency distribution relating to pixel values of a plurality of pixels in the entire scanned image.
The first discrimination method is determined as the optimum method on condition that it is determined that the distance between the peak of the outer region of the original document and the peak of the original document region is larger than a certain degree.
An image processing apparatus, characterized in that the second determination method is determined as the optimum method on condition that it is determined that the distance between the out-of-manuscript region peak and the original document region peak is smaller than a certain degree. It is an image processing device
An analytical means for analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen, and an analysis means.
Based on the analysis result by the analysis means, from among a plurality of discrimination methods for discriminating between the manuscript area which is the region where the manuscript exists and the region outside the manuscript which is the region where the manuscript does not exist in the image. An area determination means for determining an optimum method, which is the most suitable method for the scanned image, and using the determined optimum method to discriminate between the original area and the outer area.
With
The plurality of discrimination methods include a first discrimination method and a fourth discrimination method.
In the first discrimination method, among a plurality of peaks existing in the frequency distribution relating to the pixel values of a plurality of pixels in the entire scanned image, the peak having the smallest pixel value at the peak position corresponds to the out-of-manuscript region. It is determined as the peak outside the original region, and the residual peak is determined as the peak in the original region corresponding to the original region, and the pixel value of the peak outside the original and the pixel value of the peak in the original region. It is a method of discriminating between the manuscript area and the out-of-manuscript area based on the threshold value.
The fourth discrimination method is
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates in the document region, and the document is based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the region from the extra-manuscript region, the positional relationship and geometry of the detection positions of the remaining orthogonal points. Based on the relationship, the positions of at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is used as the original document. A method of discriminating as an area and determining an area other than the closed area as the extra document area,
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold value to extract edge candidates in the original area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value. Is specified as a regular edge portion of the original document area, a portion other than the straight line portion is specified as a false detection portion, and complementary processing based on the regular edge portion is performed to correct the false detection portion and correct it. It is one of the methods of discriminating the closed region formed by the later edge candidate as the document region and determining the region other than the closed region as the outer region of the document.
The analysis means determines whether or not there is a gradation of a predetermined degree or more in the entire scanned image based on the spatial distribution of the pixel values of the plurality of pixels in the scanned image.
The area determination means
The fourth discrimination method is determined as the optimum method on condition that a gradation of a predetermined degree or more is present in the entire scanned image.
An image processing apparatus characterized in that the first discrimination method is determined as the optimum method on condition that a gradation of a predetermined degree or more does not exist in the entire scanned image. It is an image processing device
An analytical means for analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen, and an analysis means.
Based on the analysis result by the analysis means, from among a plurality of discrimination methods for discriminating between the manuscript area which is the region where the manuscript exists and the region outside the manuscript which is the region where the manuscript does not exist in the image. An area determination means for determining an optimum method, which is the most suitable method for the scanned image, and using the determined optimum method to discriminate between the original area and the outer area.
With
The plurality of discrimination methods include a first discrimination method and a fourth discrimination method.
In the first discrimination method, among a plurality of peaks existing in the frequency distribution relating to the pixel values of a plurality of pixels in the entire scanned image, the peak having the smallest pixel value at the peak position corresponds to the out-of-manuscript region. It is estimated as the peak of the extra-manuscript region peak, and the residual peak is estimated as the peak of the original region, which is the peak corresponding to the manuscript region, and the pixel value of the peak of the extra-manuscript region and the pixel value of the peak of the manuscript region. It is a method of discriminating between the manuscript area and the out-of-manuscript area based on the threshold value.
The fourth discrimination method is
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates in the document region, and the document is based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the region from the extra-manuscript region, the positional relationship and geometry of the detection positions of the remaining orthogonal points. Based on the relationship, the positions of at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is used as the original document. A method of discriminating as an area and determining an area other than the closed area as the extra document area,
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold value to extract edge candidates in the original area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value. Is specified as a regular edge portion of the original document area, a portion other than the straight line portion is specified as a false detection portion, and complementary processing based on the regular edge portion is performed to correct the false detection portion and correct it. It is one of the methods of discriminating the closed region formed by the later edge candidate as the document region and determining the region other than the closed region as the outer region of the document.
The analysis means determines whether or not the distance between the extra-manuscript region peak and the manuscript region peak is larger than a certain degree in the frequency distribution relating to the pixel values of the plurality of pixels in the entire scanned image.
The region discriminating means uses the first discriminating method as the optimum method on condition that it is determined in the frequency distribution that the distance between the out-of-manuscript region peak and the manuscript region peak is larger than a certain level. Determined as
The analysis means determines whether or not there is a gradation of a predetermined degree or more in the entire scanned image based on the spatial distribution of the pixel values of the plurality of pixels in the scanned image.
The image processing apparatus is characterized in that the area discriminating means determines the fourth discriminating method as the optimum method on condition that a gradation of a predetermined degree or more is present in the entire scanned image. It is an image processing device
An analytical means for analyzing the distribution of pixel values of a plurality of pixels in a scanned image of a document placed on a platen, and an analysis means.
Based on the analysis result by the analysis means, from among a plurality of discrimination methods for discriminating between the manuscript area which is the region where the manuscript exists and the region outside the manuscript which is the region where the manuscript does not exist in the image. An area determination means for determining an optimum method, which is the most suitable method for the scanned image, and using the determined optimum method to discriminate between the original area and the outer area.
With
The plurality of discrimination methods include a second discrimination method and a fourth discrimination method.
The second discrimination method is an input / output characteristic adjustment process for adjusting an input / output characteristic between an analog input value indicating the amount of light received by the light receiving element and a digital output value indicating an output pixel value in the scan process. The entire second scan image obtained after the input / output characteristic adjustment processing is performed to adjust the input / output characteristics so that a part of the output range in the input / output characteristics before the output characteristic adjustment processing becomes a wider output range. In the frequency distribution relating to the pixel values of a plurality of pixels in the above, the peak having the smallest pixel value at the peak position among the plurality of peaks is determined as the peak in the extra-manuscript region, which is the peak corresponding to the extra-manuscript region, and the remainder is determined. The peak is determined as a manuscript region peak which is a peak corresponding to the manuscript region, and a value between the pixel value of the out-of-manuscript region peak and the pixel value of the manuscript region peak is set as a threshold, and is based on the threshold. It is a method of discriminating between the original area and the outer area of the original.
The fourth discrimination method is
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates in the document region, and the document is based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the region from the extra-manuscript region, the positional relationship and geometry of the detection positions of the remaining orthogonal points. Based on the relationship, the positions of at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is used as the original document. A method of discriminating as an area and determining an area other than the closed area as the extra document area,
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold value to extract edge candidates in the original area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value. Is specified as a regular edge portion of the original document area, a portion other than the straight line portion is specified as a false detection portion, and complementary processing based on the regular edge portion is performed to correct the false detection portion and correct it. It is one of the methods of discriminating the closed region formed by the later edge candidate as the document region and determining the region other than the closed region as the outer region of the document.
The analysis means determines whether or not there is a gradation of a predetermined degree or more in the entire scanned image based on the spatial distribution of the pixel values of the plurality of pixels in the scanned image.
The area determination means
The fourth discrimination method is determined as the optimum method on condition that a gradation of a predetermined degree or more is present in the entire scanned image.
An image processing apparatus characterized in that the second discrimination method is determined as the optimum method on condition that a gradation of a predetermined degree or more does not exist in the entire scanned image. On the computer
a) Steps to analyze the distribution of pixel values of multiple pixels in the scanned image of the document placed on the platen, and
b) Based on the analysis result in step a), a plurality of discrimination methods for discriminating between a manuscript area in which the manuscript exists and a non-manuscript area in which the manuscript does not exist in the image. From the steps, the step of determining the optimum method, which is the most suitable method for the scanned image, and
c) A step of discriminating between the manuscript area and the non-manuscript area using the optimum method determined in step b).
It is a program to execute
The plurality of discrimination methods include a first discrimination method and a third discrimination method.
In the first discrimination method, among a plurality of peaks existing in the frequency distribution relating to the pixel values of a plurality of pixels in the entire scanned image, the peak having the smallest pixel value at the peak position corresponds to the out-of-manuscript region. It is estimated as the peak of the extra-manuscript region peak, and the residual peak is estimated as the peak of the original region, which is the peak corresponding to the manuscript region, and the pixel value of the peak of the extra-manuscript region and the pixel value of the peak of the manuscript region. It is a method of discriminating between the manuscript area and the out-of-manuscript area based on the threshold value.
In the third discrimination method, the scanned image is divided into one of the vertical direction and the horizontal direction and divided into a plurality of divided regions, and the frequency distribution for each divided region is within the divided region. Of the plurality of peaks existing in the frequency distribution relating to the pixel values of the plurality of pixels, the value between the pixel value of the out-of-manuscript region peak and the pixel value of the manuscript region peak is set as the threshold value for each division region, and for each division region. It is a method of discriminating between the original area and the outer area of the original for each divided area based on the threshold value.
In step a),
a-1) A step of determining whether or not there is a gradation of a predetermined degree or more in the entire scanned image based on the spatial distribution of pixel values of a plurality of pixels in the scanned image.
Have,
In step b),
b-1) A step of determining the third discrimination method as the optimum method on condition that a gradation of a predetermined degree or more is present in the entire scanned image.
b-2) A step of determining the first discrimination method as the optimum method on condition that there is no gradation of the predetermined degree or more in the entire scanned image.
A program characterized by having. In the program of claim 18.
In step a),
a-2) When the scanned image has a gradation of the predetermined degree or more, the step of further determining the change direction of the gradation of the predetermined degree or more based on the spatial distribution situation.
Have more
In step b-1), it is a condition that it is determined that the gradation of the predetermined degree or more exists in the entire scanned image and the change direction is either the vertical direction or the horizontal direction in the scanned image. The program is characterized in that the third discrimination method is determined as the optimum method, and the change direction of the gradation of a predetermined degree or more is determined as the one direction in the third discrimination method. In the program of claim 19.
The plurality of discrimination methods also include a fourth discrimination method.
The fourth discrimination method is
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates in the document region, and the document is based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the region from the extra-manuscript region, the positional relationship and geometry of the detection positions of the remaining orthogonal points. Based on the relationship, the positions of at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is used as the original document. A method of discriminating as an area and determining an area other than the closed area as the extra document area,
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold value to extract edge candidates in the original area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value. Is specified as a regular edge portion of the original document area, a portion other than the straight line portion is specified as a false detection portion, and complementary processing based on the regular edge portion is performed to correct the false detection portion and correct it. It is one of the methods of discriminating the closed region formed by the later edge candidate as the document region and determining the region other than the closed region as the outer region of the document.
In step b),
b-3) The fourth discrimination method is the optimum method on condition that it is determined that the gradation of the predetermined degree or more is present in the entire scanned image and the change direction is the oblique direction in the scanned image. Steps to determine as,
A program characterized by having more. On the computer
a) Steps to analyze the distribution of pixel values of multiple pixels in the scanned image of the document placed on the platen, and
b) Based on the analysis result in step a), a plurality of discrimination methods for discriminating between a manuscript area in which the manuscript exists and a non-manuscript area in which the manuscript does not exist in the image. From the steps, the step of determining the optimum method, which is the most suitable method for the scanned image, and
c) A step of discriminating between the manuscript area and the non-manuscript area using the optimum method determined in step b).
It is a program to execute
The plurality of discrimination methods include a first discrimination method and a third discrimination method.
In the first discrimination method, among a plurality of peaks existing in the frequency distribution relating to the pixel values of a plurality of pixels in the entire scanned image, the peak having the smallest pixel value at the peak position corresponds to the out-of-manuscript region. It is estimated as the peak of the extra-manuscript region peak, and the residual peak is estimated as the peak of the original region, which is the peak corresponding to the manuscript region, and the pixel value of the peak of the extra-manuscript region and the pixel value of the peak of the manuscript region. It is a method of discriminating between the manuscript area and the out-of-manuscript area based on the threshold value.
In the third discrimination method, the scanned image is divided into one of the vertical direction and the horizontal direction and divided into a plurality of divided regions, and the frequency distribution for each divided region is within the divided region. Of the plurality of peaks existing in the frequency distribution relating to the pixel values of the plurality of pixels, the value between the pixel value of the out-of-manuscript region peak and the pixel value of the manuscript region peak is set as the threshold value for each division region, and for each division region. It is a method of discriminating between the original area and the outer area of the original for each divided area based on the threshold value.
In step a),
a-1) A step of determining whether or not the distance between the extra-manuscript region peak and the manuscript region peak is larger than a certain degree in the frequency distribution relating to the pixel values of a plurality of pixels in the entire scanned image.
a-2) A step of determining whether or not there is a gradation of a predetermined degree or more in the entire scanned image based on the spatial distribution of pixel values of a plurality of pixels in the scanned image.
Have,
In step b),
b-1) The first discrimination method is determined as the optimum method on the condition that it is determined in the frequency distribution that the distance between the outer-manuscript region peak and the manuscript region peak is larger than the certain degree. Steps to do and
b-2) A step of determining the third discrimination method as the optimum method on condition that a gradation of a predetermined degree or more is present in the entire scanned image.
A program characterized by having. In the program of claim 21,
The plurality of discrimination methods also include a second discrimination method.
The second discrimination method is
This is an input / output characteristic adjustment process that adjusts the input / output characteristics between the analog input value that indicates the amount of light received by the light receiving element in the scan process and the digital output value that indicates the output pixel value, and is the input before the input / output characteristic adjustment process. Pixel values of a plurality of pixels in the entire second scan image obtained after the input / output characteristic adjustment processing for adjusting the input / output characteristics so that a part of the output range in the output characteristics becomes a wider output range. In the frequency distribution with respect to, the value between the newly obtained pixel value of the extra-manuscript region peak and the newly obtained pixel value of the manuscript region peak is set as a threshold, and the manuscript region and the outside of the manuscript are based on the threshold. The method of distinguishing from the area and
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates in the document region, and the document is based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the region from the extra-manuscript region, the positional relationship and geometry of the detection positions of the remaining orthogonal points. Based on the relationship, the positions of at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is used as the original document. A method of discriminating as an area and determining an area other than the closed area as the extra document area,
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold value to extract edge candidates in the original area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value. Is specified as a regular edge portion of the original document area, a portion other than the straight line portion is specified as a false detection portion, and complementary processing based on the regular edge portion is performed to correct the false detection portion and correct it. It is one of the methods of discriminating the closed region formed by the later edge candidate as the document region and determining the region other than the closed region as the outer region of the document.
In step b),
b-3) A step of determining the second discrimination method as the optimum method on condition that it is determined that the distance between the outer-manuscript region peak and the manuscript region peak is smaller than a certain degree.
A program characterized by having more. In the program of claim 21,
In step a),
a-3) When the scanned image has a gradation of the predetermined degree or more, the step of further determining the change direction of the gradation of the predetermined degree or more based on the spatial distribution situation.
Have more
In step b-2), it is a condition that it is determined that the gradation of the predetermined degree or more exists in the entire scanned image and the change direction is either the vertical direction or the horizontal direction in the scanned image. The program is characterized in that the third discrimination method is determined as the optimum method, and the change direction of the gradation of a predetermined degree or more is determined as the one direction in the third discrimination method. In the program of claim 23
The plurality of discrimination methods also include a fourth discrimination method.
The fourth discrimination method is
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates in the document region, and the document is based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the region from the extra-manuscript region, the positional relationship and geometry of the detection positions of the remaining orthogonal points. Based on the relationship, the positions of at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is used as the original document. A method of discriminating as an area and determining an area other than the closed area as the extra document area,
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold value to extract edge candidates in the original area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value. Is specified as a regular edge portion of the original document area, a portion other than the straight line portion is specified as a false detection portion, and complementary processing based on the regular edge portion is performed to correct the false detection portion and correct it. It is one of the methods of discriminating the closed region formed by the later edge candidate as the document region and determining the region other than the closed region as the outer region of the document.
In step b),
b-4) The fourth discrimination method is the optimum method on condition that it is determined that the gradation of the predetermined degree or more is present in the entire scanned image and the change direction is the oblique direction in the scanned image. Steps to determine as,
A program characterized by having more. In the program of claim 24
The plurality of discrimination methods also include a second discrimination method.
The second discrimination method is
This is an input / output characteristic adjustment process that adjusts the input / output characteristics between the analog input value that indicates the amount of light received by the light receiving element in the scan process and the digital output value that indicates the output pixel value, and is the input before the input / output characteristic adjustment process. Pixel values of a plurality of pixels in the entire second scan image obtained after the input / output characteristic adjustment processing for adjusting the input / output characteristics so that a part of the output range in the output characteristics becomes a wider output range. In the frequency distribution with respect to, the value between the newly obtained pixel value of the extra-manuscript region peak and the newly obtained pixel value of the manuscript region peak is set as a threshold, and the manuscript region and the outside of the manuscript are based on the threshold. The method of distinguishing from the area and
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates in the document region, and the document is based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the region from the extra-manuscript region, the positional relationship and geometry of the detection positions of the remaining orthogonal points. Based on the relationship, the positions of at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is used as the original document. A method of discriminating as an area and determining an area other than the closed area as the extra document area,
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold value to extract edge candidates in the original area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value. Is specified as a regular edge portion of the original document area, a portion other than the straight line portion is specified as a false detection portion, and complementary processing based on the regular edge portion is performed to correct the false detection portion and correct it. It is one of the methods of discriminating the closed region formed by the later edge candidate as the document region and determining the region other than the closed region as the outer region of the document.
In step b),
b-5) A step of determining the second discrimination method as the optimum method on condition that it is determined that the distance between the outer document region peak and the document region peak is smaller than a certain degree.
A program characterized by having more. In the program of claim 25
The program characterized in that the second discrimination method is the same method as the fourth discrimination method. In the program of claim 25
The program characterized in that the second discrimination method is a method different from the fourth discrimination method. On the computer
a) Steps to analyze the distribution of pixel values of multiple pixels in the scanned image of the document placed on the platen, and
b) Based on the analysis result in step a), a plurality of discrimination methods for discriminating between a manuscript area in which the manuscript exists and a non-manuscript area in which the manuscript does not exist in the image. From the steps, the step of determining the optimum method, which is the most suitable method for the scanned image, and
c) A step of discriminating between the manuscript area and the non-manuscript area using the optimum method determined in step b).
It is a program to execute
The plurality of discrimination methods include a third discrimination method and a fourth discrimination method.
In the third discrimination method, the scanned image is divided into one of the vertical direction and the horizontal direction and divided into a plurality of divided regions, and the frequency distribution for each divided region is within the divided region. Of the plurality of peaks existing in the frequency distribution relating to the pixel values of a plurality of pixels, the pixel value of the extra-manuscript region peak, which is the peak corresponding to the extra-manuscript region, and the pixel value of the manuscript region peak, which is the peak corresponding to the manuscript region. It is a method of discriminating between the original area and the outer area of the original for each divided area based on the threshold value for each divided area as the value between and.
The fourth discrimination method is
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates in the document region, and the document is based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the region from the extra-manuscript region, the positional relationship and geometry of the detection positions of the remaining orthogonal points. Based on the relationship, the positions of at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is used as the original document. A method of discriminating as an area and determining an area other than the closed area as the extra document area,
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold value to extract edge candidates in the original area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value. Is specified as a regular edge portion of the original document area, a portion other than the straight line portion is specified as a false detection portion, and complementary processing based on the regular edge portion is performed to correct the false detection portion and correct it. It is one of the methods of discriminating the closed region formed by the later edge candidate as the document region and determining the region other than the closed region as the outer region of the document.
In step a),
a-1) A step of determining whether or not there is a gradation of a predetermined degree or more in the entire scanned image based on the spatial distribution of pixel values of a plurality of pixels in the scanned image.
a-2) When the scanned image has a gradation of the predetermined degree or more, a step of further determining the change direction of the gradation of the predetermined degree or more based on the spatial distribution situation, and
Have,
In step b),
b-1) The first condition is that it is determined that there is a gradation of a predetermined degree or more in the entire scanned image and the change direction is either the vertical direction or the horizontal direction in the scanned image. A step of determining the three discrimination methods as the optimum method and determining the change direction of the gradation of a predetermined degree or more as the one direction in the third discrimination method.
b-2) The fourth discrimination method is the optimum method on condition that it is determined that the gradation of the predetermined degree or more is present in the entire scanned image and the change direction is the oblique direction in the scanned image. And the steps to decide as
A program characterized by having. In the program according to any one of claims 18 to 28.
In the third discrimination method, the scanned image is divided into one of the vertical direction and the horizontal direction and divided into a plurality of divided regions, and two or more of the plurality of divided regions are used. With respect to two or more divided regions in which the frequency distributions related to the pixel values of the plurality of pixels in each divided region have two or more peaks, among the plurality of peaks existing in each frequency distribution, the two or more divided regions are related. A common position peak, which is a peak having its peak position at a common position in a frequency distribution of two or more, is determined as the document region peak, and a specific non-common position among non-common position peaks other than the common position peak is determined. The peak is determined as the extra-manuscript region peak, and the value between the pixel value of the extra-manuscript region peak and the pixel value of the manuscript region peak is set as the threshold value for each division region, and is based on the threshold value for each division region. A program characterized in that it is a method of discriminating between the document area and the area outside the document for each divided area. In the program of claim 29
In the third discrimination method, the threshold value of the divided region in which the frequency distribution regarding the pixel values of the plurality of pixels does not have a peak of 2 or more is obtained based on the threshold value of the other divided regions. A program characterized by that. On the computer
a) Steps to analyze the distribution of pixel values of multiple pixels in the scanned image of the document placed on the platen, and
b) Based on the analysis result in step a), a plurality of discrimination methods for discriminating between a manuscript area in which the manuscript exists and a non-manuscript area in which the manuscript does not exist in the image. From the steps, the step of determining the optimum method, which is the most suitable method for the scanned image, and
c) A step of discriminating between the manuscript area and the non-manuscript area using the optimum method determined in step b).
It is a program to execute
The plurality of discrimination methods include a first discrimination method and a second discrimination method.
In the first discrimination method, among a plurality of peaks existing in the frequency distribution relating to the pixel values of a plurality of pixels in the entire scanned image, the peak having the smallest pixel value at the peak position corresponds to the out-of-manuscript region. It is estimated as the peak of the extra-manuscript region peak, and the residual peak is estimated as the peak of the original region, which is the peak corresponding to the manuscript region, and the pixel value of the peak of the extra-manuscript region and the pixel value of the peak of the manuscript region. It is a method of discriminating between the manuscript area and the out-of-manuscript area based on the threshold value.
The second discrimination method is
This is an input / output characteristic adjustment process that adjusts the input / output characteristics between the analog input value that indicates the amount of light received by the light receiving element in the scan process and the digital output value that indicates the output pixel value, and is the input before the input / output characteristic adjustment process. Pixel values of a plurality of pixels in the entire second scan image obtained after the input / output characteristic adjustment processing for adjusting the input / output characteristics so that a part of the output range in the output characteristics becomes a wider output range. In the frequency distribution with respect to, the value between the newly obtained pixel value of the extra-manuscript region peak and the newly obtained pixel value of the manuscript region peak is set as a threshold, and the manuscript region and the outside of the manuscript are based on the threshold. The method of distinguishing from the area and
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates in the document region, and the document is based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the region from the extra-manuscript region, the positional relationship and geometry of the detection positions of the remaining orthogonal points. Based on the relationship, the positions of at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is used as the original document. A method of discriminating as an area and determining an area other than the closed area as the extra document area,
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold value to extract edge candidates in the original area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value. Is specified as a regular edge portion of the original document area, a portion other than the straight line portion is specified as a false detection portion, and complementary processing based on the regular edge portion is performed to correct the false detection portion and correct it. It is one of the methods of discriminating the closed region formed by the later edge candidate as the document region and determining the region other than the closed region as the outer region of the document.
In step a),
a-1) A step of determining whether or not the distance between the out-of-manuscript region peak and the manuscript region peak is larger than a certain degree in the frequency distribution relating to the pixel values of a plurality of pixels in the entire scanned image.
Have,
In step b),
b-1) A step of determining the first determination method as the optimum method on condition that it is determined that the distance between the out-of-manuscript region peak and the original document region peak is larger than a certain degree.
b-2) A step of determining the second discrimination method as the optimum method on condition that it is determined that the distance between the outer-manuscript region peak and the manuscript region peak is smaller than a certain degree.
A program characterized by having. On the computer
a) Steps to analyze the distribution of pixel values of multiple pixels in the scanned image of the document placed on the platen, and
b) Based on the analysis result in step a), a plurality of discrimination methods for discriminating between a manuscript area in which the manuscript exists and a non-manuscript area in which the manuscript does not exist in the image. From the steps, the step of determining the optimum method, which is the most suitable method for the scanned image, and
c) A step of discriminating between the manuscript area and the non-manuscript area using the optimum method determined in step b).
It is a program to execute
The plurality of discrimination methods include a first discrimination method and a fourth discrimination method.
In the first discrimination method, among a plurality of peaks existing in the frequency distribution relating to the pixel values of a plurality of pixels in the entire scanned image, the peak having the smallest pixel value at the peak position corresponds to the out-of-manuscript region. It is estimated as the peak of the extra-manuscript region peak, and the residual peak is estimated as the peak of the original region, which is the peak corresponding to the manuscript region, and the pixel value of the peak of the extra-manuscript region and the pixel value of the peak of the manuscript region. It is a method of discriminating between the manuscript area and the out-of-manuscript area based on the threshold value.
The fourth discrimination method is
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates in the document region, and the document is based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the region from the extra-manuscript region, the positional relationship and geometry of the detection positions of the remaining orthogonal points. Based on the relationship, the positions of at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is used as the original document. A method of discriminating as an area and determining an area other than the closed area as the extra document area,
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold value to extract edge candidates in the original area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value. Is specified as a regular edge portion of the original document area, a portion other than the straight line portion is specified as a false detection portion, and complementary processing based on the regular edge portion is performed to correct the false detection portion and correct it. It is one of the methods of discriminating the closed region formed by the later edge candidate as the document region and determining the region other than the closed region as the outer region of the document.
In step a),
a-1) A step of determining whether or not there is a gradation of a predetermined degree or more in the entire scanned image based on the spatial distribution of pixel values of a plurality of pixels in the scanned image.
Have,
In step b),
b-1) A step of determining the fourth discrimination method as the optimum method on condition that a gradation of a predetermined degree or more is present in the entire scanned image.
b-2) A step of determining the first discrimination method as the optimum method on condition that there is no gradation of the predetermined degree or more in the entire scanned image.
A program characterized by having. On the computer
a) Steps to analyze the distribution of pixel values of multiple pixels in the scanned image of the document placed on the platen, and
b) Based on the analysis result in step a), a plurality of discrimination methods for discriminating between a manuscript area in which the manuscript exists and a non-manuscript area in which the manuscript does not exist in the image. From the steps, the step of determining the optimum method, which is the most suitable method for the scanned image, and
c) A step of discriminating between the manuscript area and the non-manuscript area using the optimum method determined in step b).
It is a program to execute
The plurality of discrimination methods include a first discrimination method and a fourth discrimination method.
In the first discrimination method, among a plurality of peaks existing in the frequency distribution relating to the pixel values of a plurality of pixels in the entire scanned image, the peak having the smallest pixel value at the peak position corresponds to the out-of-manuscript region. It is estimated as the peak of the extra-manuscript region peak, and the residual peak is estimated as the peak of the original region, which is the peak corresponding to the manuscript region, and the pixel value of the peak of the extra-manuscript region and the pixel value of the peak of the manuscript region. It is a method of discriminating between the manuscript area and the out-of-manuscript area based on the threshold value.
The fourth discrimination method is
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates in the document region, and the document is based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the region from the extra-manuscript region, the positional relationship and geometry of the detection positions of the remaining orthogonal points. Based on the relationship, the positions of at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is used as the original document. A method of discriminating as an area and determining an area other than the closed area as the extra document area,
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold value to extract edge candidates in the original area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value. Is specified as a regular edge portion of the original document area, a portion other than the straight line portion is specified as a false detection portion, and complementary processing based on the regular edge portion is performed to correct the false detection portion and correct it. It is one of the methods of discriminating the closed region formed by the later edge candidate as the document region and determining the region other than the closed region as the outer region of the document.
In step a),
a-1) A step of determining whether or not the distance between the extra-manuscript region peak and the manuscript region peak is larger than a certain degree in the frequency distribution relating to the pixel values of a plurality of pixels in the entire scanned image.
a-2) A step of determining whether or not there is a gradation of a predetermined degree or more in the entire scanned image based on the spatial distribution of pixel values of a plurality of pixels in the scanned image.
Have,
In step b),
b-1) The first discrimination method is determined as the optimum method on the condition that it is determined in the frequency distribution that the distance between the outer-manuscript region peak and the manuscript region peak is larger than the certain degree. Steps to do and
b-2) A step of determining the fourth discrimination method as the optimum method on condition that a gradation of a predetermined degree or more is present in the entire scanned image.
A program characterized by having. On the computer
a) Steps to analyze the distribution of pixel values of multiple pixels in the scanned image of the document placed on the platen, and
b) Based on the analysis result in step a), a plurality of discrimination methods for discriminating between a manuscript area in which the manuscript exists and a non-manuscript area in which the manuscript does not exist in the image. From the steps, the step of determining the optimum method, which is the most suitable method for the scanned image, and
c) A step of discriminating between the manuscript area and the non-manuscript area using the optimum method determined in step b).
It is a program to execute
The plurality of discrimination methods include a second discrimination method and a fourth discrimination method.
The second discrimination method is
This is an input / output characteristic adjustment process that adjusts the input / output characteristics between the analog input value that indicates the amount of light received by the light receiving element in the scan process and the digital output value that indicates the output pixel value, and is the input before the input / output characteristic adjustment process. Pixel values of a plurality of pixels in the entire second scan image obtained after the input / output characteristic adjustment processing for adjusting the input / output characteristics so that a part of the output range in the output characteristics becomes a wider output range. Of the plurality of peaks, the peak having the smallest pixel value at the peak position is determined as the extra-manuscript region peak, which is the peak corresponding to the extra-manuscript region, and the remaining peak is set in the manuscript region. It is determined as the original area peak which is the corresponding peak, and the value between the pixel value of the extra document area peak and the pixel value of the original document area peak is set as a threshold, and the original area and the outside of the original are based on the threshold. It is a method to distinguish from the area,
The fourth discrimination method is
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold to extract edge candidates in the document region, and the document is based on the positions of four orthogonal points where the edge candidates are orthogonal to each other. When it is determined that at least one of the four orthogonal points is not detected in distinguishing the region from the extra-manuscript region, the positional relationship and geometry of the detection positions of the remaining orthogonal points. Based on the relationship, the positions of at least one orthogonal point are estimated so that the closed region formed by connecting the four orthogonal points with a straight line has a rectangular shape, and the closed region is used as the original document. A method of discriminating as an area and determining an area other than the closed area as the extra document area,
The scanned image is subjected to binarization processing and edge extraction processing based on a provisional threshold value to extract edge candidates in the original area, and among the extracted edge candidates, a straight line portion having a length equal to or longer than a predetermined value. Is specified as a regular edge portion of the original document area, a portion other than the straight line portion is specified as a false detection portion, and complementary processing based on the regular edge portion is performed to correct the false detection portion and correct it. It is one of the methods of discriminating the closed region formed by the later edge candidate as the document region and determining the region other than the closed region as the outer region of the document.
In step a),
a-1) A step of determining whether or not there is a gradation of a predetermined degree or more in the entire scanned image based on the spatial distribution of pixel values of a plurality of pixels in the scanned image.
Have,
In step b),
b-1) A step of determining the fourth discrimination method as the optimum method on condition that a gradation of a predetermined degree or more is present in the entire scanned image.
b-2) A step of determining the second discrimination method as the optimum method on condition that there is no gradation of the predetermined degree or more in the entire scanned image.
A program characterized by having.

Images