JP3052464B2 - Contour point extraction method using multi-valued data - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for extracting contour points using multi-valued data, and more particularly to a method for extracting contour points using multi-valued data for easily obtaining a smooth contour with high accuracy.

[0002]

2. Description of the Related Art In recent years, digital processing of image information has been performed using an information processing apparatus including a computer and the like. By the way, when processing image information such as characters and figures, if information processing is performed in pixel units, a large amount of memory resources are consumed because the amount of data to be processed is large. It is unsuitable for performing image processing such as enlargement, reduction, rotation, etc. of an image, and such an image processing may cause a figure to be distorted or crushed. And other problems. Therefore, an outline vector is obtained from information on the outline shape of an image such as a character or a figure, and the character or figure is represented by a limited number of outline vectors, thereby enlarging or reducing the image with a small amount of image information. Many proposals have been made to enable free image processing such as rotation and rotation.

By the way, image processing for obtaining an outline vector from image data such as characters and figures obtained by reading a manuscript having a gradation number of 2 with an image reading apparatus in the presence of characters and figures in the background. In the related art, it has been performed to obtain an outline vector according to a procedure as shown in FIG. 10, for example. In FIG. 10, 7
Is a document in which characters, figures, and the like to be subjected to image processing are described. Image information (image data) of the characters and figures described in the document 7 is stored in a block 8 in FIG. The image data is read into the image processing apparatus by using an appropriate image information reading apparatus as shown in (1), for example, an image scanner or a television camera capable of outputting image information as digital data. The digital data of the image information read into the image processing device is subjected to a binarization process by the binarization processing unit 9 so that the background pixel and the character or graphic pixel can be distinguished well. After that, the data of the contour point is extracted using the point where the background pixel and the character or graphic pixel are adjacent in block 10 as the contour point. Is determined, and a line segment sequence expressing a contour line that follows is determined, and the contour is vectorized.

Here, the vectorization of the contour described above with reference to the arrangement of contour points (contour point sequence) shown in FIG. 11 will be described. First, a large number of squares shown in FIG. Assuming that each object is a contour point, a contour point having an X coordinate of 5 and a Y coordinate of 3 and an X coordinate of 9 and a Y
A line segment connecting the contour point with coordinates 3 is a vector of one contour, and also connects a contour point with an X coordinate of 9 and a Y coordinate of 3 and a contour point with an X coordinate of 9 and a Y coordinate of 7 A line segment is a vector of another contour, and a line segment connecting a contour point having an X coordinate of 9 and a Y coordinate of 7 with a contour point having an X coordinate of 5 and a Y coordinate of 7 is another one. A line connecting a contour point having an X coordinate of 5 and a Y coordinate of 7 and a contour point having an X coordinate of 5 and a Y coordinate of 3 is another contour vector. The arrangement of a large number of contour points illustrated in FIG. 11 is indicated by the four contour vectors corresponding to the above-described four line segment strings, and the block 1
As shown in FIG. 8, vectorized contour data is output.

FIG. 12A shows the digital data of the image information read by the image processing apparatus, which is subjected to the binarization processing by the binarization processing section 9 as described with reference to FIG. FIG. 1 is a diagram showing the above-mentioned binarized data as a display image corresponding to the binarized data so that the state of the binarized data obtained can be easily understood visually.
2 (b) is a display image corresponding to the contour point data corresponding to the contour point data so that the state of the contour point data extracted based on the binarized data can be easily understood visually. Further, FIG. 12 (c) shows a contour obtained by determining a line segment sequence expressing a contour tracing the contour using the data of the contour points described above and performing vectorization of the contour. FIG. 5 is a diagram showing the contour vector data as a display image corresponding thereto so that the state of the vector data can be easily understood visually.

[0006]

As described above with reference to FIG. 10, a character obtained by reading an original having two tones with an image reading device and having a character, a figure, etc. in the background. When image processing such as obtaining contour point data and contour vectors from image data such as images and figures is performed,
The digital data read into the image processing device using an image reading device capable of outputting image information as digital data undergoes a binarization process so that the background pixel and the character or figure pixel can be distinguished well. After that, the data of the contour point is extracted with the point where the background pixel and the character or graphic pixel are adjacent to each other as the contour point, and then the contour line following the contour is expressed using the above-mentioned contour point data. Image processing is performed such that a line segment sequence is determined and contour vectorization is performed. However, as described above, the original to be read by the image reading device has characters or characters in the background. Even if a document has a figure or the like and the number of gradations is 2, the image data such as characters and figures read by the image reading device is not binary digital data but multi-valued digital data. You have me.

Image data obtained by reading an original having characters and figures in the background and having two gradations by an image reading apparatus is not binary digital data but multi-valued digital data. This and the problems caused by the above points will be described below with reference to FIGS. Many small squares indicated by broken lines in FIG. 13A indicate the size of the photoelectric conversion element for each pixel in the photoelectric conversion unit of the image reading device. By the way, as shown in FIG. 13A, the image information which is subjected to the photoelectric conversion by the photoelectric conversion element for each pixel having a certain size is such that the number of gradations is two. Even in the case of image information of a document, the amount of light given to the individual photoelectric conversion elements described above is the amount of light corresponding to the white portion of the image, and the amount of light corresponding to the black portion of the image. Not only in the case of, the light amount of the white part of the image and the light amount of the black part of the image may be given to the photoelectric conversion element, and in this case, the original does not originally exist from the photoelectric conversion element in the document. An electric signal corresponding to the halftone is generated, and halftone digital data is output.

FIG. 13B shows a case where the amount of light in the state illustrated in FIG. 13A is given to the photoelectric conversion element for each pixel of the photoelectric conversion unit. FIG. 14B is a diagram showing digital data obtained in correspondence with an electric signal generated by a photoelectric conversion element for each pixel of the photoelectric conversion unit. In FIG. 13B, white circles (white circles) indicate white levels. Represents a pixel from which digital data can be obtained, and a circle with a hatched circle (hereinafter referred to as a hatched circle) represents a pixel from which digital data of a black level can be obtained. A large number of dots and circles (hereinafter referred to as dotted circles) represent pixels from which halftone level digital data can be obtained. The meanings of the above-mentioned white circles, hatched circles, dotted circles and the like are the same in other drawings. FIG. 14 shows a case where a character or a figure is read from a black pixel indicating a background in image data read by an image reading apparatus on a document in which white characters or figures are present in a black background and the number of gradations is 2. FIG. 9 illustrates a state of distribution of the number of data for each gradation up to the white pixel shown in FIG.
“b” indicates the boundary between the halftone area and the background black area, and “Thw” indicates the boundary between the halftone area and the area indicating the characters and figures.

The digital data output from the reading device due to the above-mentioned cause includes digital data corresponding to halftone image information which did not originally exist in the original document. The state of the binarized signal, that is, the state of the contour point, depends on the set value of the threshold set in the above-described binarization processing performed on the digital data of the image information read in Although the position may change, this point is shown in FIGS.
The following is a description with reference to FIG. First, FIG.
(A) shows the density distribution of the paper surface in a document in which characters, graphics, etc. are present in the background and the number of gradations is two, as indicated by the dotted line in FIG. Further, the solid line shown as a reading result in the drawing is a document obtained by reading an original having a density distribution indicated by a dotted line indicating the density distribution on the paper surface and having a gradation number of 2 with an image reading device. 9 illustrates an example of the obtained reading result.

The example of the read result indicated by the curve shown in FIG. 15A is shown at a boundary portion between a white portion and a black portion in a document which originally has only a binary density distribution. Because one photoelectric conversion element in the photoelectric conversion unit of the image reading device was located, the portion corresponding to the boundary between the white portion and the black portion in the document did not originally exist in the document. It shows the state of output data when halftone output data is being output from the reader. The dashed-dotted line Th shown in FIG.
1, Th2 indicates two examples of thresholds used when binarizing the output data of the reader illustrated by the curve shown by the solid line. (B) of FIG. 15 is binarized data obtained by binarizing the read result indicated by the curve in (a) of FIG. 15 with a threshold indicated by a dashed line Th1; FIG. 15C shows the result of reading indicated by the curve in FIG.
This is binarized data obtained by binarizing with a threshold indicated by a one-dot chain line Th2.
As described above, when halftone data is included in the output data from the reading device, the state of the binarized signal, that is, the position of the contour point changes depending on the way of setting the threshold for binarization. To happen.

FIG. 16 shows that the position of a contour point changes depending on how to set a threshold for binarization when halftone data is included in output data from a reading device. It is a figure for explaining. FIG. 16A is a diagram showing the state of the density distribution of the graphic on the paper surface of the document, and FIG. 16F is a diagram of the graphic illustrated in FIG. 16A. It is a figure showing the outline figure expected as an outline. In FIG.
16 (b) and (d) show the approximate density distribution of the figure of the original in FIG.
(A), the original is read by the image reader as described above, and the output data of the image reader is binarized at different thresholds. FIGS. 16 (b) and 16 (d) illustrate the states of the two binarized data in order to make it easier to visually understand the states of the binarized data. The above-described binarized data is shown as a display image corresponding to the binarized data. The binarized data illustrated in FIG. 16B is, for example, a threshold T.
h1 is obtained as a result of the binarization process, and FIG.
The binarized data illustrated in (d) of FIG. 6 is obtained as a result of, for example, binarization processing with a threshold Th2.

When the approximate density distribution of the original figure is as shown in FIG. 16A, output data obtained by reading it with an image reading device is converted into binary data at different thresholds. The reason why the two binarized data obtained by the conversion may be completely different from each other as illustrated in, for example, FIGS. Has irregularities in the order of pixel size, uneven density, or the presence of spots, etc.
This is because halftone data is generated when the image is read by the image reading device as described above. The outline (array of outline points... Outline point sequence) extracted from the binarized data illustrated in (b) of FIG.
The outline (array of outline points... Outline point sequence) extracted from the binarized data illustrated in (d) of FIG. e)
It is as illustrated in FIG.

As described above, the threshold set for the binarization process by the digital data corresponding to the halftone image information which does not originally exist in the original and is generated in the digital data output from the reading device. The state of the binarized signal, that is, the position of the contour point changes in units of the size of the pixel depending on the setting value of, and it is conventionally difficult to obtain a good contour line. There was a need for a solution.

[0014]

According to the present invention, multi-valued data obtained by reading a document having characters and figures in the background and having two gradations by an image reading device is converted into white level pixels. Means for distinguishing between data, pixel data of a black level, and pixel data of each of different intermediate levels in N (where N is a number of 1 or more) intermediate levels, and the set number of intermediate levels N For the numerical value N determined correspondingly,
Means for setting the coordinate system so that the pixel is divided into (N + 1) in the vertical direction and the horizontal direction, and corresponding to the state where the white level pixel and the black level pixel are in contact with each other,
Means for setting an outline point between the white level pixel and the black level pixel, and, when there is a halftone pixel, a relative gradation between pixels around the halftone pixel The level relationship is examined, and the halftone pixel is equally divided vertically and horizontally (N + 1) in correspondence with the relative tone level relationship between the halftone pixel and the peripheral pixels. Means for setting a predetermined position in a position as a contour point position, and providing a contour point extraction method using multi-valued data.

[0015]

The multi-valued data obtained by reading an original having a number of gradations of 2 in which characters, figures, and the like are present in the background by an image reading apparatus are converted into white level pixel data, black level pixel data, and N levels. Pixel data for each of different intermediate levels (where N is one or more) is distinguished.
Numerical value N determined corresponding to the set number N of intermediate levels
, The pixels are (N + 1) in the vertical and horizontal directions, respectively.
The coordinate system is set so as to be divided. Corresponding to the state where the white level pixel and the black level pixel are in contact with each other, an outline point is set between the above white level pixel and the black level pixel, and when there is a halftone pixel Is
The relative gradation level relationship between the halftone pixel and the peripheral pixels is examined, and the relative gradation level relationship between the halftone pixel and the peripheral pixels is determined. Then, a predetermined position is set as a contour point position among the positions equally divided (N + 1) vertically and horizontally in the halftone pixel.

[0016]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram of a contour point extracting method using multivalued data according to the present invention. FIG. 1 is a block diagram used for a general description of an outline point extraction method using multilevel data according to the present invention.
In FIG. 1, 1 is an image data source, 2 is a multi-value processing unit, 3 is a halftone pixel extraction unit, 4 is a contour point determination unit,
Reference numeral 5 denotes a contour point deletion unit surrounded by contour points, and 6 denotes a contour point data output unit. The contour point extraction method using multi-valued data of the present invention is based on multi-valued data obtained by reading a document having characters and graphics in the background and having two gradations by an image reading device. Means for distinguishing between white level pixel data, black level pixel data, and N (where N is 1 or more) intermediate level pixel data for each different intermediate level, and the set intermediate level Means for setting a coordinate system such that a pixel is divided into (N + 1) pixels in a vertical direction and a horizontal direction, respectively, for a numerical value N determined in correspondence with the number N of white pixels and black level pixels Means for setting a contour point in the middle between the white level pixel and the black level pixel in correspondence with the state in which Of relative gradation level between pixels It examined, corresponding to the relative gray level of the relationship between the above-mentioned halftone pixel and the surrounding pixels, vertical in the halftone pixel, the horizontal (N
+1) means for setting a predetermined position as a contour point position among the equally divided positions. In the following description, when N is 1,
That is, the coordinate system is set so that the pixel data is divided into white level pixel data, black level pixel data, and one intermediate level pixel data, and the pixel is divided into two in the vertical direction and the horizontal direction. In this embodiment, the number of image data to be output is 2 × 2 = 4 times the number of input image data. In this embodiment, eight projections are connected and four depressions are connected. The contour line formed by the outputted contour points is smoother than the contour line formed by the contour points determined after binarization as in the conventional method, and the sequence of the contour points is simple. It has a closed loop and has no mustaches or shared points.

The image data read by an image reading apparatus on an original having characters and figures in the background and having two gradations is not multi-valued digital data but multi-valued digital data. In FIGS. 13 (a) and 13 (b) in FIG. 13 (b), which are used for the description of The pixel for which halftone level digital data is obtained is sandwiched between the pixel for which white level digital data is indicated by a white circle and the pixel for which black level digital data is indicated by a hatched circle. However, as in the case described above, when an image reading apparatus reads a document in which characters, graphics, and the like are present in the background and the number of gradations is 2, the halftone level is changed. When digital data is obtained, image information targeted for photoelectric conversion is given to a photoelectric conversion element for each pixel having a certain size in a state as shown in FIG. However, since the individual photoelectric conversion elements are provided with the light amount of the white part of the image and the light amount of the black part of the image, an electric signal corresponding to the halftone is generated, thereby Key digital data is output.

As described above, halftone level digital data can be obtained when an image reading apparatus reads a document in which characters, graphics, and the like are present in the background and has two gradations. 13A and 13B and FIG. 13C and FIG. 13D, the multivalued data of the present invention is used. The contour point extraction method described above uses a method in which a halftone level is generated in a state as described with reference to FIGS. 13A and 13B, that is, a pixel from which halftone level digital data is obtained is: This is well applied only when the pixel is located between a pixel at which white-level digital data is obtained and a pixel at which black-level digital data is obtained.
If a halftone level is generated in the state described with respect to (d), the contour point extraction method using the multi-valued data of the present invention is not applied. {(C) and (d) of FIG. When a halftone level is generated in the state described above, not only the direction of the contour and the number of contours are not easily determined, but also information at a high frequency is lost at the time of sampling. There is no point in pursuing it because it is gone.

FIG. 2 is a diagram showing a basic example of how to take a contour in the contour point extraction method using multi-valued data according to the present invention. FIG. FIG. 2B is a symbol of a pixel for which it is impossible to determine whether a pixel constitutes a character or a figure or a pixel which constitutes a background.
(C) is a symbol showing a contour point and a contour line. When a pixel constituting a character or a figure and a pixel constituting a background are adjacent to each other, the contour line is shown in FIG. As shown in (d), the light passes between pixels constituting a character or a figure and pixels constituting a background. The number of contour points in the figure is larger than the original number of pixels because, for the numerical value 1 determined in correspondence with the number N of intermediate levels described above, the pixels in the vertical and horizontal directions are respectively ( This is because the coordinate system is set so as to be divided into (N + 1) = 2, and the possible points of the contour are both at the center of the pixel and at the point between the pixels. . In addition, the contour of a pixel which cannot be determined as a pixel constituting a character or a figure or a pixel constituting a background passes through the center of the pixel as shown in FIGS. 2 (e) and 2 (f).

FIG. 3 shows four pixels Pu, Pd, Pl, and Pr in a four-connected relationship with the halftone level pixels indicated by the dotted circles in FIG. In the case where a pixel of a halftone level indicated by a dotted circle shown in FIG. 3A is a pixel having a gradation as shown in FIG. 3B, (a) of FIG. The pixel of the halftone level indicated by the dotted circle in ()) is adopted as the pixel of the halftone level as shown in FIG. 3C, and the halftone dot indicated by the dotted circle in FIG. FIG. 3 shows four pixels Pu, Pd, Pl, and Pr in a four-connected relationship with the tonal level pixels.
If the pixel of the halftone level indicated by the dotted circle shown in FIG. 3A is not a pixel having the gradation as shown in FIG. 3B, the pixel shown in FIG. FIG. 4 is a diagram for explaining that a pixel of a halftone level indicated by a dot circle is not adopted as a pixel of a halftone level as shown in FIG. As described above, the pixel data of the halftone level pixel that is not adopted as the halftone level pixel as shown in FIG.
By binarizing at an appropriate threshold, the pixel data is changed to either pixel data of white level or pixel data of black level. Therefore, as an example of the case where the contour point extraction method using the multi-valued data of the present invention is not applied, a pixel of a halftone level is generated in the state described with reference to FIGS. 13C and 13D. The case will be excluded.

As shown in FIG. 2 described above, a contour point occurs at the center of a pixel, between two pixels, or where four pixels are in contact. Whether or not a certain coordinate position becomes a contour point is determined only by pixels that include or touch the coordinate position. That is, by paying attention to a certain pixel P in FIG. 4 and examining three pixels, that is, a right pixel Pr, a lower pixel Pd, and a lower right pixel Prd of the noted pixel P, a small black circle in FIG. It is possible to determine whether or not each of the four contour point candidates indicated by the mark can be a contour point. Therefore, M ×
For the entire screen having N pixels, the same processing is sequentially performed on (M−1) × (N−1) pixels other than the rightmost and lowermost pixels, so that the contour is obtained. Points can be determined. Since the number of contour points is four per pixel of interest, the total number is 4 × (M−1) × (N−1).

FIG. 5 is a diagram showing conditions for evaluating each contour point candidate and determining whether or not to be a contour point. FIG. 6 shows the conditions for the contour points in FIG. Conditions for deleting the outline points performed by the deletion unit 5 for the enclosed outline points,
That is, it is a diagram illustrating a condition that “contour points surrounded by contour points on the upper, lower, left, and right sides are deleted”. Hereinafter, the evaluation of each contour point candidate and the condition for determining whether or not to be a contour point will be described with reference to FIG. First, (1) a case where the center of a pixel is a contour point. When a pixel is at a halftone level, a point at the center of a pixel of interest becomes a contour point by taking the idea that a contour passes through the center. Is the case where the target pixel is at the halftone level and is limited to that case.
(2) When a point between two pixels is an outline point: A point between the two pixels is a point between the target pixel and the right pixel and a point between the target pixel and the lower pixel. Two points with the point in between. Considering that a contour line passes between a white level pixel and a black level pixel, a point between the white level pixel and the black level pixel is a contour point. When one of the two pixels is a pixel of a halftone level,
Since the contour passes through the center of the halftone pixel,
In this case, the point between the two pixels does not become a contour point. When both of the two pixels are pixels at the halftone level, the contour points are used to connect the center contour points of the two pixels at the halftone level.

(3) A case in which a point adjacent to four pixels is an outline point: A point adjacent to four pixels does not include both a background pixel and a character or figure pixel in the four pixels. In this case, it is not considered as a contour point. However, even when the background pixel and the character or figure pixel are included in the four pixels, the contour point is considered in consideration of the connection of the contour line. May not. The rules described above are (1) and (2)
Since the number of pixels at the halftone level is 0, 1, and 2 out of the four pixels, it is considered that the number of pixels is slightly more complicated than in the case of. In the case where the number of pixels at the halftone level is three or four out of the four pixels, both the background pixel and the character or graphic pixel are not included. You don't need to think about First, in the case where the number of halftone level pixels is 0 out of the four pixels, the background pixel region and the character or graphic pixel region are
When it is separated vertically or horizontally, it is a contour point.
Next, among the four pixels, the number of pixels at the halftone level is 0.
When two background pixels are present diagonally and two characters or graphics pixels are also present diagonally, contour points are formed at the top, bottom, left and right. Not a point.

When the number of halftone level pixels among the four pixels is 0, the number of pixels of the background is one and the number of pixels of characters and figures is three. In the case where the figure or the figure has one pixel, the manner in which the contour points are connected differs depending on whether or not it is adopted as a contour point. In other words, if it is adopted as a contour point, it will be 4-connected, and if it is not adopted as a contour point, it will be 8-connected. If the pixel area of a character or a figure is connected by eight at the convex part and four by the concave part, the contour point can be tracked by a simple algorithm. A good result can be obtained if the number of pixels is three and the contour point is not. If the number of background pixels is three and the number of characters or graphics is one, it is not a contour point.

In the case where the number of pixels of the halftone level is one of the four pixels, if there is a pixel in the same level state next to the pixel of the halftone level in the vertical direction and the horizontal direction, that pixel is The region extends in the oblique direction, and it is considered that the region of the pixel at the diagonal to the pixel of the halftone level is divided by the region extending in the oblique direction, so that the outline candidate point does not become the outline point. In cases other than the above, contour points are used as contour points to ensure the connectivity of contour lines.

In the case where the number of pixels at the halftone level is two out of the four pixels, if the two pixels at the halftone level are diagonal, the two halftone level levels Are defined as contour points to connect the centers of the pixels of When two halftone level pixels are arranged in the vertical or horizontal direction, a contour line connecting the centers of the two halftone level pixels, a background pixel, and a character or graphic In order to connect the contour lines extending between the pixels, the contour candidate points are contour points. At this time, the way in which the outline extends is not uniquely determined, but this point is solved by applying the conditions described above with reference to FIG.

When considering a point adjacent to a certain point, as shown in FIG.
Conventionally, four connections considering only one point and eight connections considering eight adjacent points in each of the vertical and horizontal oblique directions as shown in FIG. 7B have been considered. FIG. 8 shows image data provided by an arrangement of black pixels (hatched circle pixels) and white pixels (open circle pixels) as shown in FIGS. 8A and 8D, respectively. FIG. 8 is a diagram showing how contour points are determined when the number of connected contour lines is four and eight when the contour line is connected. FIG. 8B
FIG. 8A shows contour points when the number of connected contour lines is four in the image data shown in FIG. 8A, and FIG. 8C shows the contour points in FIG. 8 shows contour points when the number of connected contour lines is eight in the image data shown in FIG. 8, and FIG.
8 (d) shows the outline points when the number of connected outlines is 4 in the image data shown in FIG. 8 (d), and FIG. 8 (f) shows the outline points in FIG. 8 (d). 3 shows contour points when the number of connected contour lines is eight for the image data.

The contour points extracted by applying the contour point extraction method using the multi-valued data of the present invention are as follows. , And the concave portion becomes four-connected. (2) The contour point does not stretch like a beard with one point width. (3) The outline points do not touch each other across the background pixel area (there is no shared point)
, The contour points can be easily tracked. That is, the extraction of the contour points by the contour point extraction method using the multi-valued data of the present invention is performed as follows in the image processing apparatus illustrated in FIG. ... The bitmap data of the contour point is scanned from the upper left to the lower right with priority in the horizontal direction. When the contour point is found next to the background pixel, the process goes to. Go to when the contour point is found next to the pixel or contour point of the character or figure. ... Adopted as the start point of the contour line, set the used flag, and set the right direction as the direction data. (1): 45 counterclockwise from the opposite direction of the direction data
The contour points are searched in order from the eight connection points in the turned direction. If an unused contour point is found, it is adopted as a point where the contour line extends, and the used flag is set as the direction data and the used flag is set, and (1) is repeated. If a used contour point is not found, the last contour point is adopted as the end of the contour line, and the procedure goes to (2).

(2) The last contour point is adopted as the end of the contour line, and the used flag is set. ... Adopted as the start point of the contour line, set the used flag, and set the right direction as the direction data. (1): Contour points are searched in order from eight connection points in a direction rotated clockwise by 45 degrees from the opposite direction of the direction data. If an unused contour point is found, it is adopted as a point where the contour line extends,
The used flag is set as the direction data using the found direction, and (1) is repeated. If a used contour point is not found, go to (2). (2) adopts the last contour point as the end of the contour line, sets the used flag, and goes to FIG. 9 illustrates the above-described contour tracking state. In FIG. 9, hatched circles indicate unused contour points, double circles indicate used contour points, and FIG. FIG. 9B is a diagram illustrating scanning for finding a contour point, FIG. 9B is a diagram illustrating tracking of a contour point in a counterclockwise direction,
FIG. 9C is a diagram illustrating the tracking of the contour points in the clockwise direction, and FIG. 9D is a diagram illustrating the last unused contour points.

FIG. 17 shows a manuscript used by the present inventor for the experiment of the contour point extraction method using the multi-valued data of the present invention, that is, a black background (in FIG. 17, a black background is indicated by hatching. FIG. 18 shows an example of an outline point extraction method using multi-valued data according to the present invention. The contour points were extracted using the image of FIG. 18A showing the experimental results when image processing was performed by the used contour point extraction method, and the binarized data from the same original as in the above case. FIG. 19 is a diagram illustrating a result of image processing according to a conventional method in comparison with an image illustrated in FIG. FIG. 19 shows a case where image processing is performed by a contour point extraction method using ternary data as an embodiment of the contour point extraction method using multi-value data according to the present invention, similarly to the case of FIG. 18 described above. (A) of FIG. 19 showing the experimental results of FIG. 19 and (b) of FIG. 19 showing the results of image processing according to the conventional method in which contour points were extracted using the binarized data from the same original as in the above case. It is the figure which showed b) and the image of (c) in comparison. The image of FIG. 19 (c) is twice as large as the image of FIG. 19 (a) by enlarging the image of FIG. 19 (b) twice for comparison with the image of FIG. 19 (a). It is a thing that has been decided. Further, FIG. 20 shows an experimental result when image processing is performed by a contour point extraction method using ternary data as an embodiment of the contour point extraction method using multi-valued data of the present invention.
0 (a) and the same original 2
A part of the image is shown in an enlarged manner so that the comparison result with the image of FIG. 20B showing the result of the image processing by the conventional method in which the contour points are extracted using the quantified data can be easily understood. FIG. Comparison between the experimental result when image processing is performed by the contour point extraction method using the multi-valued data of the present invention and the result of image processing by the conventional method in which contour points are extracted using the binarized data 18 to FIG.
Looking at each figure in FIG. 20, it can be seen that the contour point extraction method using the multi-valued data of the present invention can obtain better contour points than the conventional method.

[0031]

As is apparent from the above description, the contour point extraction method using multi-valued data according to the present invention has a gradation number of 2 in which characters, figures, etc. exist in the background. Multi-valued data obtained by reading the original document by the image reading apparatus is divided into white level pixel data, black level pixel data and N intermediate levels (where N is 1 or more) at different intermediate levels. Pixel data. For a numerical value N determined corresponding to the set number N of the intermediate levels, the coordinate system is set so that the pixel is divided into (N + 1) in the vertical direction and the horizontal direction, and the white level pixel and the black Corresponding to the state where the pixels of the level are in contact with each other, an outline point is set in the middle between the pixel of the white level and the pixel of the black level. The relative gradation level relationship between the pixel around the pixel and the surrounding pixel is examined, and the relative gradation level relationship between the halftone pixel and the surrounding pixel is compared with the halftone pixel. Since the predetermined position is set as the contour point position in the position of the pixel equally divided vertically and horizontally (N + 1), the contour point is extracted using the conventional binary data. The size of the unevenness is 1 / (N +
1), a highly accurate image can be easily obtained,
The above-mentioned conventional problems can be solved well.

[Brief description of the drawings]

FIG. 1 is a block diagram used for schematically describing the whole outline point extraction method using multi-valued data according to the present invention.

FIG. 2 is a diagram showing a basic example of how to take a contour in a contour point extraction method using multi-valued data of the present invention.

FIG. 3 is a diagram for explaining the adoption and non-employment of pixels of a halftone level.

FIG. 4 is a diagram showing conditions for determining whether a contour point candidate can be a transit point.

FIG. 5 is a diagram showing conditions for evaluating each contour point candidate to determine whether to be a contour point.

FIG. 6 is a diagram illustrating conditions for deleting a contour point performed by a contour point deletion unit surrounded by the contour points in FIG. 1;

FIG. 7 is an explanatory diagram of four connections and eight connections.

FIG. 8 shows a case where image data is arranged in an arrangement of black pixels (hatched circle pixels) and white pixels (open circle pixels), and the number of connected contour lines is four and eight. It is a figure showing how a contour point is determined, respectively.

FIG. 9 is a diagram for explaining tracking of a contour line.

FIG. 10 is a block diagram of an image processing apparatus that obtains an outline vector from image data.

FIG. 11 is an explanatory diagram of a contour point array and a contour point vector.

FIG. 12 is an explanatory diagram showing a correspondence among a document, a contour point array, and a contour point vector.

FIG. 13 is a diagram illustrating a relationship between a photoelectric conversion element and image information in a photoelectric conversion unit of the image reading device.

FIG. 14 is a diagram showing a gradation distribution of read image data.

FIG. 15 is an explanatory diagram showing that a difference occurs in pixel data due to a difference in threshold.

FIG. 16 is an explanatory diagram showing that a difference occurs in a contour due to a difference in threshold.

FIG. 17 is a diagram showing an original for an experiment.

FIG. 18 shows an experimental result in the case where image processing is performed by a contour point extraction method using ternary data as an embodiment of the contour point extraction method using multi-valued data according to the present invention; FIG. 11 is a diagram illustrating a result of image processing according to a conventional method in which contour points are extracted by using binary data of the same document.

FIG. 19 shows an experimental result when image processing is performed by a contour point extraction method using ternary data as an embodiment of the contour point extraction method using multi-valued data according to the present invention, and FIG. FIG. 11 is a diagram illustrating a result of image processing according to a conventional method in which contour points are extracted by using binary data of the same document.

FIG. 20 shows an experimental result in the case where image processing is performed by a contour point extraction method using ternary data as an embodiment of the contour point extraction method using multi-valued data according to the present invention; FIG. 11 is a diagram illustrating a result of image processing according to a conventional method in which contour points are extracted by using binary data of the same document.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Image data source, 2 ... Multi-value processing part, 3 ... Half tone pixel extraction part, 4 ... Contour point determination part, 5 ... Contour point deletion part surrounded by contour points, 6 ... Contour point data Output section,

Claims (2)

(57) [Claims] 1. A multi-level data obtained by reading a document having a gradation number of 2 with an image reading apparatus in which a character, a graphic, or the like is present in a background is converted into white level pixel data and black level pixel data. Means for discriminating between N (where N is a number equal to or greater than 1) intermediate level pixel data for each different intermediate level, and a numerical value N determined corresponding to the set number N of intermediate levels. Means for setting the coordinate system so that the pixel is divided into (N + 1) in the vertical direction and the horizontal direction, and corresponding to the state where the white level pixel and the black level pixel are in contact with each other, Means for setting an outline point between the white level pixel and the black level pixel,
When there is a halftone pixel, the relative gradation level relationship between the halftone pixel and the surrounding pixels is checked, and the relative tone level between the halftone pixel and the surrounding pixels is determined. Means for setting a predetermined position as a contour point position among positions equally divided vertically and horizontally (N + 1) in a halftone pixel in correspondence with the relationship of the various gradation levels. Contour point extraction method using digitized data. 2. The contour point extraction method according to claim 1, wherein the value of N is set to 1.