KR20080095743A - Image processing program and image processing device - Google Patents

Abstract

An image processing program and an image processing device are provided to extract a character, a stamp mark, a mark, etc. in high density from a documental image having color distortion and dim colors. An image obtaining module(101) obtains a color image or a brightness image from a paper document through a scanner or an OCR(Optical Character Reader), and a background removal data generating module(102) generates background removal data from the color image or the brightness image inputted to an CPU(Central Processing Unit). A contour color conversion data generating module(103) generates contour line color conversion data in which a contour color of an input character, a ruled line or a digit line is converted into a color of an contour inner part and to which the color image and the background removal data are inputted. A specific object discrimination module(104) generates data, which indicates an input character part based on the contour line conversion data as to the background removal data inputted to the CPU, and outputs a result as to a specific object discrimination.

Description

Image processing program and image processing device {IMAGE PROCESSING PROGRAM AND IMAGE PROCESSING DEVICE}

According to the present invention, for example, a document such as a journal is photographed by an optical character reader (OCR: Optical Reader Reader), a scanner, a digital camera, or the like, and a specific object such as written characters, prints, marks, etc. is extracted from the generated document image. An image processing method and an image processing apparatus.

In financial institutions and local governments, scanners such as OCR are used to increase the efficiency of document processing tasks such as forms. The main functions of the OCR are the generation of document images, extraction of characters in the document images, and character recognition. As a kind of generated document image, there are a binary image, a luminance image, and a color image.

In the process using the binary image, since the amount of data is small, the processing time is reduced. However, in binary image processing, it is difficult to distinguish between ruled lines, digit lines, guide letters, hatchings, and prewritten letters, which are pre-printed on a form, and when written characters that are written by hand or printed later greatly overlap. Therefore, there is a problem that noise may occur in the extraction result of the character, or a part of the extracted character may be missing, resulting in incorrect character recognition.

The processing using the luminance image is black and white tint image processing. In the luminance image processing, since the luminance values of the preprinted and written characters are distinguished from each other by using different ones, when the preprinted and written characters overlap, it is easier to distinguish them from the binary image processing. However, when the preprinting and the luminance values of the written characters are close, their discrimination accuracy is lowered.

In the process using a color image, pre-printing and written characters can be distinguished by color difference, so that distinguishing them becomes easier than in luminance image processing. In color image processing, writing characters, printing, marks, and the like are extracted by dropping out the color of pre-printing.

The method includes dropping out the designated dropout color before entering the document as shown in "Patent Document 3", and extracting a particular shape portion such as a ruled line in the entered document as "Patent Document 1" or "Patent Document 2". And dropping out the same color as that of the extracted part.

[Patent Document 1] Japanese Patent Laid-Open No. 2003-196592

[Patent Document 2] Japanese Patent Laid-Open No. 2005-258683

[Patent Document 3] Japanese Patent Laid-Open No. 2006-134355

[Patent Document 4] Japanese Patent Laid-Open No. 2004-336106

[Patent Document 5] Japanese Patent Laid-Open No. 2005-18810

In the color image processing, color shift occurs in an image generated from an OCR, a scanner, or a digital camera, so that a specific object such as a written character or human face cannot be extracted correctly, and a part of the pre-printing remains or a part of the specific object. There is a problem that is missing.

The color shift means that the position of at least one of the color components of the three primary colors sensed, the R value of the red component, the G value of the green component, and the B value of the blue component are shifted. As a cause of color shift, the chromatic aberration of a lens, the arrangement position of a sensor, a conveyance speed, etc. are mentioned, for example. In particular, in a scanner using a two-dimensional CCD such as a stand-type scanner or a digital camera, color shift due to chromatic aberration occurs a lot.

Due to the color shift, a false color different from the original color of the specific object is generated in the outline portion of the specific object such as pre-printing or writing character. For example, in the outline of a black letter, there exist a case where the red and blue false color generate | occur | produce, or the case where the light red false color generate | occur | produces in the outline of a blue ruled line. Therefore, the color information causes an error in color image processing for distinguishing written characters from preprinting.

In contrast, Patent Document 4 attempts to eliminate chromatic aberration of the lens, and Patent Document 5 eliminates color shift due to the arrangement position of the sensor.

However, even if the method of measuring and correcting the shift amount is performed, it is difficult to completely remove the color shift from the image. In addition, more accurate color shift correction also causes a problem that requires a long time for ruled lines.

In addition, in the color image processing and the luminance image processing, when a color blur occurs in an image, a specific object such as a writing character or a printout cannot be extracted correctly, and a part of the pre-printing is left or a part of the character is missing. There is a problem.

With color blur, the color of the ruled line and the outline of the writing character is blurred, and light color is generated. Color blurring increases the dispersion of color information such as red component, blue component, red component, brightness, saturation, color, and luminance of preprinted and written characters, making it difficult to distinguish between written characters and preprinted text.

SUMMARY OF THE INVENTION The present invention has been made in view of these problems, and an object of the present invention is to provide an image processing method and an image processing apparatus for accurately extracting a specific object such as a writing character, printing, mark, etc. from a document image including color shift or color blurring. have.

In order to achieve the above object, the present invention is an image processing method for extracting a specific object such as a written character, a print, a mark, or the like from a color image or a luminance image obtained by reading a document such as a book with a scanner or a digital camera. Or background removal data generation processing for removing a background from a luminance image and generating background removal data indicating a portion other than the background, and a background of color information of an outline other than the background in a color image or a portion other than the background in the luminance image. It is characterized by having the outline color conversion data generation process which produces | generates the data converted into the color information of the pixel inside the outline of other part, and the specific object extraction process which extracts a specific object part.

Further, the contour color conversion data generation process refers to the neighboring pixels which are a plurality of pixels in the vicinity of the pixel of interest in the color image or the luminance original image, and the red component, blue component, red component, and brightness of the pixel of interest. And color information such as saturation, color, and luminance to generate low luminance color expansion luminance data converted into color information of the pixel having the lowest luminance value among the neighboring pixels and the pixel of interest.

The specific object discrimination processing is characterized in that ruled line extraction, specific object candidate extraction, estimation of color information of ruled line and color information of specific object, and determination of specific object are performed.

According to the present invention, even a color image or luminance image with color shift or color blur, it is possible to accurately distinguish certain objects such as pre-printing, writing characters, printing, marks, and the like, and to extract only writing characters with high precision. Done. It is possible not only to write characters but also to extract specific objects in a document image, such as printing and a mark, with high precision.

EMBODIMENT OF THE INVENTION Hereinafter, the image processing method and image processing apparatus to which this invention is applied are demonstrated in detail using drawing.

Example 1

2 is a diagram illustrating an embodiment of the image processing apparatus of the present invention.

This includes the communication device 201, the image acquisition device 202, the display device 203, the external memory device 204, the memory 205, the CPU (Central Processing Unit) 206, and an input device such as a keyboard or a mouse. 207 is an image processing apparatus 208 connected by a communication line such as a PCI bus.

The program having the configuration of the specific object extraction processing shown in FIG. 1 is stored in a storage device such as the external storage device 204 or the memory 205 and executed by the CPU 205.

The color image and luminance image of the document input to the CPU may be input from an image acquisition device 202 or a communication device 201 such as a scanner or an OCR, or may be stored in the external storage device 204.

The result of the specific object extraction processing may be output to the display device 203, output to the outside via the communication device 201, or use in other programs in the image acquisition processing device 208. An example of another program is a program that performs character recognition.

FIG. 27 shows an example in which a color image input from the image acquisition device 202 or the communication device 201 or stored in the external storage device 204 is displayed on the display window 2701 on the display device 203. . 28 is an example which displayed the result of the specific object extraction process on the display window 2702 on the display 203.

1 is a diagram illustrating a configuration of a specific object extraction processing program to which the present invention is applied. The specific object extraction processing program is composed of an image acquisition module 101, a background removal data generation module 102, an outline color conversion data generation module 103, a specific object determination module 104, and a control module 105. .

The image acquisition module 101 performs image acquisition processing for acquiring a color image or luminance image obtained by imaging a paper document or the like by a scanner, an OCR, or the like.

The background removal data generation module 102 performs background removal data generation processing for generating background removal data from a color image or a luminance image input to the CPU 206.

For example, when the color image including the ruled line 301, the digit line 302, and the preprinting and writing characters 304 of the hatching 303 as shown in FIG. 3 is acquired, the background removal data generation module 102 And data indicating ruled lines, digit lines, and written character portions shown in FIG. 4.

Background removal data generation processing is a process of removing the background part in an image, and extracting the preprinted part and writing character part of a ruled line and a digit line. There are various methods for realizing this, but the method shown in FIG.

First, in the luminance value data generation processing 601, a luminance image represented by luminance is generated from a color image represented by three primary colors (R value, G value, and B value) of RGB. Next, in the block division process 602, the luminance image is divided into a plurality of blocks. Finally, in the binarization processing 603, binary data is generated for each block in which a pixel having a low luminance value is black and a pixel having a high luminance value is white in a block. The binary data generated by this is background removal data in which black pixels represent portions other than the background, as shown in FIG.

The outline color conversion data generation module 103 receives the color image 604 and the background removal data 605, and converts the outline color of the ruled line, the digit line, and the write character into the color of the inner part of the outline. The outline color conversion data generation process of generating the conversion data 1303 is performed. In addition, the color image 604 may be a luminance image.

The specific object discrimination module 104 generates data indicating the written character portion shown in FIG. 5 with reference to the contour color conversion data 1303 for the background removal data 605 input to the CPU 206. The specific object determination process is performed, and the specific object discrimination result 706 is output.

Here, with reference to Fig. 7, the conventional specific object discrimination process will be described. In the conventional specific object discrimination process, background removal data is input, and the determination result of the specific object was output with reference to the color image.

The conventional specific object discrimination process is shown in FIG. First, in a ruled line extraction process 701 ', a ruled line portion is extracted. In this processing, the ruled line portion is extracted by extracting the portion where the black pixels in the background removal data are long and linearly connected. The result is FIG.

Next, in ruled line removal processing 702 ', ruled line removal data from which the ruled line portion is removed from the background elimination data is generated. The result is shown in FIG.

Next, in the specific object candidate extraction process 703 ', the writing character part candidate which becomes a candidate of the writing character part used as a specific object is extracted from the ruled line removal data using the information of the size and position of a rectangle. The result is shown in FIG.

Next, in the estimation process 704 'of the ruled color and the specific target color, the reference character candidate partial color that is the ruled line partial color information that is the color information of the ruled line portion and the color information of the write character candidate portion by referring to the color image 604 Estimate the information.

Next, in the determination processing 705 ′ of the specific object, using the ruled partial color information and the write character candidate partial color information, it is determined whether each pixel of the black pixel portion in the background removal data is a pixel of the write character. do. This processing is processing for determining whether the color information of the color image belongs to the ruled partial color information or the writing character candidate partial color information at the position of each pixel of the black pixel portion in the background removal data.

Specifically, the following processing is performed for each pixel of the black pixel portion in the background removal data 605. In the processing of arbitrary black pixel positions (Xa, Xb) in the background removal data 605, the color information in (Xa, Xb) of the color image 604 is estimated by the ruled color and the specific target color (704 '). It determines whether it is close to the ruled partial color information or the written character candidate partial color information outputted by the " If the color information of (Xa, Xb) is close to the ruled partial color information, the position of (Xa, Xb) is the ruled line, and if the color information of (Xa, Xb) is close to the write character candidate partial color information, It is determined that the position of (Xa, Xb) is the write character portion.

As the color information of the color image 604, the R, G, and B values of the RGB three primary colors may be used, or color information obtained by converting them, for example, the luminance value, the hue, saturation, and brightness of the HSV space may be used. . In addition, only one value may be used among them and a plurality of values may be used. In addition, various discrimination algorithms using teacher data can be used for the discrimination method. For example, neural networks, linear identifiers, Maharanobis distances, etc. are used.

Next, with reference to the color image 604, the specific object determination process is performed and the specific object discrimination process is complete | finished by outputting the specific object discrimination result 706 '.

However, in the conventional specific object discrimination process, if there is a color shift in the input color image 604, the color information in the color image 604 is also shifted, so that either the ruled portion color information or the write character candidate partial color information The determination itself based on the color information of whether or not there is a possibility may be shifted, and there existed a problem that it was impossible to distinguish a preprinted and written character by color information. For this reason, in the specific object discrimination result obtained by the conventional method, there existed the output of the discrimination result of this invention, for example, as shown in FIG.

Here, FIG. 11 is an example of the image (writing character) with color shift. In FIG. 11, blue and red false colors are generated in the outline of black written characters.

12 is an example of an image (rulered line) with color shift. In Fig. 12, a light red false color is generated in the outline of the blue ruled line. Consider a case where only the write character is extracted using color information from an image including black write characters and blue ruled lines.

If there is no color shift between the written character and the ruled line, only the written character can be extracted in the conventional specific object discrimination process. However, in the image with the color shift as shown in Figs. 11 and 12, since the red component exists in either of the outline of the writing character and the outline of the ruled line, the case where the outline of the ruled line occurs as noise or part of the character is missing. There is something to be done. When color shift occurs in this way, there is a problem that preprinting and writing characters cannot be distinguished by color information.

11 and 12, in the image processing apparatus to which the present invention is applied, contour color conversion data generation processing is performed, and specific object discrimination processing is performed with reference to the contour color conversion data.

The above-mentioned outline color conversion data generation module 103 performs outline color conversion data generation processing. Specifically, in the portion other than the background in the color image 604, data obtained by converting the color of the outline of the portion other than the background into the color of the pixel inside the outline of the portion other than the background is generated. That is, data obtained by converting the color information of the ruled line, the number of digits, and the outline of the write character into the color information of the inner part of the outline is generated.

13 is an example of a specific processing flow of contour color conversion data generation processing.

In the outline color conversion data generation processing, the color image 604 and the background removal data 605 are input from the communication device 201, the image acquisition device 202, or the external storage device 204 through the memory 205. .

In the near luminance value generation processing 1301 and the low luminance color expansion processing 1302, pixels in regions other than the background in the color image are selected one by one (the selected pixels are called the pixel of interest), and the color of the pixel of interest is selected. The information is converted. These two processes are repeated until all the pixels in the area other than the background in the color image have been processed.

In the neighborhood luminance value generation processing 1302, luminance values of pixels in a field surrounding the vicinity of the pixel of interest (in the example of FIG. 14, 9 pixels in a field in a 3x3 range centering on the pixel of interest) are respectively generated. do. Hereinafter, pixels other than the pixel of interest in the field are called neighboring pixels. The field is not limited to 3x3, but may be 2x2 or 4x4, for example. Note that the pixel of interest is not limited to the center in the field, but the field may be set to be located at any one position in the field.

Next, in the low luminance color expansion process 1302, the color information (for example, the R value, the G value, and the B value) of the pixel of interest is converted into the color information of the pixel having the lowest luminance value among the pixel of interest and the neighboring pixel. . This converts the color information of the contour portion in which the R value, the G value, and the B value are shifted to the color information inside the contour portion, and is a process of pseudoly converting the false color into the original color information.

More specifically, the luminance values of the pixel of interest and the neighboring pixels in the field are calculated, the pixel having the lowest luminance value is extracted, and the color information of the pixel of interest is converted into the color information of the pixel having the lowest luminance value. If the luminance value of the pixel of interest is the lowest luminance value, the color information of the pixel of interest is maintained as it is. In this way, outline color conversion data 1303, which is data obtained by converting color information, is generated from the ruled line, the number of digits, and the portion of the write character in the color image 604.

According to the contour color conversion data generation process, for example, in the case of the write character, as shown in FIG. 14, the red and blue false colors having high luminance in the outline portion of the write character shown in FIG. The luminance at is converted to low black.

Further, according to the contour color conversion data generation process, for example, in the case of ruled lines, as shown in FIG. 15, a light red high color having a high luminance in the outline portion of the ruled lines shown in FIG. The luminance at is converted to low blue.

16 is a diagram of a specific processing flow of the specific object discrimination processing in the first embodiment.

First, background removal data is input and ruled line extraction processing 701 for extracting ruled line portions is performed.

Next, ruled line removal processing 702 for generating ruled line removal data from which the ruled line portion is removed from the background removal data is performed.

Next, the specific object candidate extraction process 703 which extracts the writing character part candidate used as a candidate of the writing character part used as a specific object from the ruled line removal data is performed using the information of the size and position of a rectangle.

Next, in the specific object discrimination processing of the present invention, the RGB values of the contour color conversion data 1303 are referred to in the estimation process 1601 and the determination process 1602 of the ruled color and the specific object color.

In the area of the contour color conversion data 1303 corresponding to the black pixel area of the background color removal data, there are fewer pixels having a false color generated due to color shift, so that the estimation of the specific target color and the ruled color becomes more accurate, and the result is As a result, the accuracy of discriminating ruled lines and specific objects is also improved.

As described above, in the image processing apparatus 208 to which the present invention is applied, the writing character portion is made black and the ruled portion is made blue in the estimation 1602 of the ruled color and the specific target color and the determination processing 1602 of the specific target. Because of this, the written character portion can be correctly determined.

As mentioned above, since the image processing apparatus 208 refers to the contour color conversion data including the RGB value after the contour color conversion processing, it is possible to accurately extract the written character to be a specific object from the color image containing the color shift. Can be. In addition, the character recognition apparatus which takes as input the writing character extraction result which is the output of this image processing apparatus can obtain a more accurate recognition result. The extraction of the written characters is used as an example, but the extraction with the printed characters and marks can be similarly performed with high precision.

Next, another embodiment of the present invention will be described.

Example 2

As shown in Fig. 17, the specific object discrimination unit 104 may be a specific object extraction process in which the specific object is determined using only the estimation of the ruled color.

In the process shown in FIG. 17, in the process of estimating ruled line color 1701, only the color information of the ruled line is estimated with reference to the contour color conversion data. Next, in the removal processing 1702 of the ruled color portion, by using the ruled color information, the ruled color portion is removed from the background removal data 605 to determine the written character portion to be the specific object.

Example 3

As shown in Fig. 18, the specific object discrimination unit 104 may be a specific object extraction process in which the specific object discrimination process is performed using only the estimation of the specific object color.

In the processing shown in FIG. 18, in the estimation process 1801 of the specific target color, only the color information of the specific target candidate is estimated with reference to the contour color conversion data 1303. Next, in the extraction processing 1702 of the specific target color portion, the write character portion to be the specific target is extracted from the background removal data 605 using the color information of the specific target.

Example 4

As shown in Fig. 19, the specific object discrimination unit 104 may be a specific object extraction process in which specific object discrimination processing is performed using clustering.

The process shown in FIG. 19 performs determination using only the color information of the part other than a background, without using the result of ruled line extraction. First, in the clustering process 1901, the contour color conversion data 1303 of a portion other than the background is clustered. For clustering, the R, G, and B values of the RGB three primary colors may be used, or color information obtained by converting them, for example, luminance values, hue, saturation, and brightness of the HSV space may be used. In addition, only one value may be used among them and a plurality of values may be used. The clustering method includes a k-means method, a region expansion method, a discriminant analysis method, and the like.

Next, in the selection process 1902 of the cluster of a specific target, the cluster of a specific target is selected from the plurality of clusters obtained by clustering. There are various methods for selection, but there are methods such as selecting a cluster having a high luminance value.

Then, in the extraction 1901 of the specific target cluster color portion, a write character that is a specific target is extracted by extracting a pixel having the color information of the selected cluster from the black pixel portion of the background removal data.

Example 5

It is good also as a specific object extraction process which added the color shift correction module 2001 to the structure of the specific object extraction process program shown in FIG.

This specific target extraction processing program has the structure as shown in FIG. 20, and performs the same process as the embodiment shown in FIG. 1 except the process demonstrated next.

The color shift correction module 2001 performs color shift correction processing. The color shift correction process is a color shift that is data which reduced the color shift by shifting or expanding or reducing the R, G, and B values of the color image 604 obtained by the document image acquisition process. Generate calibration data.

Incidentally, in the configuration shown in Fig. 1, the data inputted to the background removal data generation process and the contour color conversion data generation process use the color image 604. In the embodiment of Fig. 20, the background removal data is used. The data input to the generation process and the outline color conversion data generation process are color shift correction data. This makes it possible to extract specific objects such as written characters with high accuracy even in an image having a large amount of color shift.

Example 6

It is good also as a specific object extraction process which added the designated color acquisition module 2101 to the structure of the specific object extraction process program shown in FIG.

This specific object extraction process has the structure as shown in FIG. 21, and performs the same process as the embodiment shown in FIG. 1 except the process demonstrated next.

The designated color acquisition module performs the specified color acquisition process. The designated color acquisition process acquires the index specified extraction object color information 2203 specified as the specific object to extract. The designated extraction target color information may be one previously designated by a user or input from an input device such as a keyboard or a mouse. The color information may be an R value, a G value, or a B value of RGB, and may be color information obtained by converting them, for example, a luminance value, a hue, saturation, and brightness of an HSV space. In addition, only one value may be used among them and a plurality of values may be used. Moreover, the value which shows one color may be sufficient, and the value which shows the range of the value of a color may be sufficient.

Then, the specific object discrimination process is a process of including the designated extraction object color information 2203 in the input as in FIG. 22 or FIG.

FIG. 22 performs determination 2201 of a specific object using the specified extraction object color information 2203 and the result of the estimation 1601 of the ruled color and the specific object color in the specific object discrimination process of FIG. 16.

FIG. 23 specifies 2301 of a cluster of a specific object using the specified extraction object color information 2203 in the specific object discrimination process using the clustering 1901 of FIG. 19.

The embodiment described so far is effective not only for the problem of color shift but also for the problem of color blur. 24 shows an example of color blurring in which the outline portion of the written character is light colored. When the outline color conversion data generation process is performed on the image of FIG. 24, the outline color conversion data 1303 shown in FIG. 25 is generated. In the outline color conversion data 1303, the inside of the color image 604 to be input is blurred, and the light color portion is converted into a dark color. Thereby, it becomes possible to extract a specific object with high precision also about the image with color blurring.

The embodiment described so far is effective not only when a color document is input but also when a luminance image in which color blur is generated is input. In the case where a luminance image is input, in the embodiment of Fig. 1, the outline color conversion data generation processing can be made by the process shown in Fig. 26.

FIG. 26 shows the luminance image 2604 and the background removal data 605, and the low luminance color expansion process 2601 of the luminance image is performed one pixel at a portion other than the background in the luminance image 2604. Goes. Then, in the parts other than the background, that is, the ruled lines, the number of digits, and the written characters, the outline color conversion data 1303, which is data obtained by converting the luminance values in the luminance image 2604, is generated.

In the low luminance color expansion process 2601 of the luminance image, the luminance value of the pixel having the lowest luminance value among the pixel of interest and the neighboring pixel is converted into the luminance value of the pixel of interest.

1 is a diagram illustrating a configuration of a specific object extraction process.

2 illustrates an image processing apparatus.

3 is an example of a color image.

4 is background removal data.

5 is a determination result of a specific target.

6 is a diagram illustrating an example of background removal data generation processing;

7 is a diagram illustrating a conventional specific object discrimination process.

8 is a ruled line extraction result.

9 is the ruled line removal result.

10 shows a specific target candidate extraction result.

11 is an example of color shift of a written character.

12 is an example of color misalignment of ruled lines.

13 is a diagram illustrating an example of contour color conversion data generation processing.

FIG. 14 is a diagram illustrating contour color conversion data generation processing of the image of FIG. 11; FIG.

FIG. 15 is a diagram illustrating contour color conversion data generation processing of the image of FIG. 12; FIG.

16 is a diagram illustrating a specific object discrimination process.

17 is a diagram showing a specific object discrimination process using only estimation of ruled color;

18 is a diagram showing a specific object discrimination process using only the estimation of the specific object color;

19 is a diagram illustrating a specific object discrimination process using clustering.

20 is a diagram illustrating a configuration of a specific target extraction processing program to which color shift correction is applied.

Fig. 21 is a diagram showing the configuration of a specific target extraction processing program with an extraction target color specifying function.

Fig. 22 is a diagram showing a species-specific object discrimination processing program including a designated extraction object color.

Fig. 23 is a diagram showing a specific object discrimination process using clustering including a specified extraction object color.

24 illustrates an example of color blurring.

Fig. 25 is a diagram showing a state of contour color conversion data generation processing for an image with color blurring;

Fig. 26 is a diagram showing an example of contour color conversion data generation processing of luminance image input.

27 is a display example of a color image.

28 is a display example of a determination result of a specific target.

<Explanation of symbols for the main parts of the drawings>

101: image acquisition module

102: background removal data generation module

103: contour color conversion data generation module

104: specific target determination module

105: control module

Claims (15)

Means for removing the background from the input image information and generating background removal data representing an area other than the background; Means for generating color conversion data obtained by converting color information of a pixel of a contour relating to a specific object into color information of a pixel inside the contour in an area other than the background of the input image information; Means for storing the background removal data and the color conversion data; Means for selecting a specific target from the background removal data and for outputting a specific target portion with reference to the color conversion data; Image processing apparatus comprising a. The method of claim 1, And the image information is color image information or luminance information of an image. The method of claim 1, In the means for generating the color conversion data, Selects a field including a plurality of pixels from background removal data representing an area other than the background, Generate luminance values of pixels in the field, And color conversion data obtained by converting color information of the pixel of interest in the field into color information of a pixel having the lowest luminance value in the pixel in the field. The method of claim 1, And the image information is any one of information stored in an external storage device, information acquired by an image acquisition device, or information input from a communication device. The method of claim 1, The means for generating the color conversion data, Setting a pixel of interest on the image information and referring to a neighboring pixel that is a pixel in a field surrounding the vicinity of the pixel of interest, And low-intensity color-expanded luminance data in which color information of the pixel of interest is converted to color information of a pixel having the lowest luminance value among neighboring pixels and pixel of interest. The method of claim 1, And the color information is at least one of red component, blue component, red component, brightness, saturation, hue, and luminance. The method of claim 1, Means for extracting ruled line information from the background removal data; Means for removing ruled line information extracted from the background removal data; Means for determining a specific target from which the specific target candidate is selected from the background removal data from which the ruled information is removed, and outputting a specific target portion with reference to the color conversion data; The image processing apparatus further comprises. By using a processing apparatus including a storage means, an image acquisition means, arithmetic means, and display means, Removing a background from the image information input from said image acquisition means, and generating background removal data representing an area other than the background; Generating color conversion data obtained by converting color information of a pixel of a contour of a specific object into color information of a pixel inside the contour, in an area other than the background of the input image information; Storing the background removal data and the color conversion data in a storage means; A process of specific object discrimination which selects a specific object candidate from the background removal data stored in said storage means, and outputs a specific object portion to said display means with reference to said color conversion data. Image processing method comprising a. The method of claim 8, And the image information is color image information or luminance information of an image. The method of claim 8, In the step of generating the color conversion data, Selecting a field including a plurality of pixels from background removal data representing image information other than the background, Generate the luminance values of the pixels in the field, And generating color conversion data in which the luminance value of the pixel of interest in the field is converted to the lowest luminance value in the pixel in the field. The method of claim 8, The image information is any one of information stored in an external storage device, information acquired by an image acquisition device, or information input from a communication device. The method of claim 8, In the step of generating the color conversion data, Regarding the pixel of interest in the field, reference is made to nearby pixels which are pixels other than the pixel of interest in the field, And low-intensity color expansion luminance data in which color information of the pixel of interest is converted into color information of a pixel having the lowest luminance value among the neighboring pixel and the pixel of interest. The method of claim 8, And the color information is at least one of a red component, a blue component, a red component, brightness, saturation, hue, and luminance. The method of claim 8, Extracting ruled line information from the background removal data; Removing ruled line information extracted from the background removal data; A process of determining a specific target by selecting a specific target candidate from the background removing data from which the ruled information is removed and referring to the color conversion data to output a specific target portion. The image processing method further comprises. Computer for image processing, Means for removing the background from the image information input from the image acquisition means and generating background removal data representing an area other than the background; Means for generating color conversion data obtained by converting color information of pixels of an outline relating to a specific object into color information of a pixel inside the outline, in an area other than the background of the input image information; Means for storing said background removal data and color conversion data in a storage means, Means for determining a specific target from the background removal data stored in the storage means, and for outputting the specific target portion to the display means with reference to the color conversion data. An image processing program for functioning as a system.

Images