JP2001351068A - Device and method for recognizing character, device and method for processing image and computer readable recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To recognize a character in a color document image by more simple constitution by utilizing the features of the color document image. SOLUTION: A character recognition device for inputting a color document image and recognizing a character in the color document image is provided with a gray convention circuit 1 for converting the maximum value or minimum value of plural color component signals in the color signal of each of pixels constituting the color document image into a gray image as a representative value of the pixel, a binarizing circuit 2 for binarizing the gray image converted by the circuit 1 and a character recognition circuit 3 for recognizing a character from the binary data obtained from the circuit 2.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character recognition process for recognizing characters by inputting a color document image or the like, a character recognition device for performing image processing, a character recognition method, an image processing device, and the like.
The present invention relates to an image processing method and a computer-readable recording medium.

[0002]

2. Description of the Related Art Conventionally, a document to be recognized by a character recognition apparatus has been generally printed in black and white, such as a newspaper or a magazine. However, in recent years, with the spread of personal computers and the like, color printers have also been spreading along with lower prices. For this reason, the document to be subjected to character recognition is often a color document image.

[0003] In addition, it has become possible to recognize color document images as they are by improving the performance of personal computers. However, color images have limitations such as an increase in memory size. Therefore, a color image is converted into a low-resolution color image, or a gray image is used as an input image, and a resolution conversion process is internally performed to obtain a high-resolution 2
It is conceivable to reduce the memory size at the time of input by creating a value image and performing character recognition on the binary image.

A technique for obtaining a binarized image by variably setting a binarization threshold using density information or density change information around a pixel of interest (the pixel in question) is disclosed in Japanese Patent Laid-Open No. 5-2 / 1993.
No. 25329.

[0005]

However, since a low-resolution image is used as an input image, an optimum high resolution is not always obtained when performing character recognition even if resolution conversion is performed. For this reason, there has been a problem that an apparatus or a processing method must be complicated to obtain an optimum high resolution.

Further, when the background density is obtained from the input color image, it is necessary to cope with a case where the background density changes. Further, in the technique disclosed in Japanese Patent Application Laid-Open No. 5-225392, since the threshold value for binarization is variably set, it is necessary to compare the pixel with the pixel after determining the threshold value. However, there was a problem that the length was longer.

SUMMARY OF THE INVENTION The present invention has been made in view of the above, and it is a first object of the present invention to recognize a character of a color document image with a simpler configuration or method by utilizing features of the color document image. And

It is a second object of the present invention to be able to cope with a change in background density by obtaining peripheral color information. A third object is to improve the processing efficiency in the binarization processing.

[0009]

According to a first aspect of the present invention, there is provided a character recognition apparatus for inputting a color document image and performing character recognition on the color document image. In the color document image, a maximum value or a minimum value of a plurality of color component signals in a color signal of each pixel constituting the color document image is converted into a gray image as a representative value of the pixel by a gray conversion unit, and converted by the gray conversion unit. And a character recognizing means for performing character recognition from the binary data of the binarizing means.

According to the present invention, a color document image, which is a target document of the character recognition device, is often printed by a printing device. Therefore, color signals other than black and white are represented by color signals (RGB, CMY (K)). ) In many cases, and the background (paper) portion of the document is often white or high in brightness. To distinguish it from the background, the brightness value of the text portion is generally low. Utilizing the characteristic of the color document image that the target pixel is a target, the maximum value or the minimum value of a plurality of color component signals in the color signal is converted into a gray image as a representative value of the pixel by the gray conversion means, and then binarized. Character recognition using the two representative values (representative colors) can be realized with a simple configuration.

Further, in the character recognition device according to the present invention, the color component signal is the same as the color component signal of the ink used when printing the document to be recognized.

According to the present invention, since a color document image is often printed by a printing apparatus, a color component signal other than black and white is set to be the same as the ink used when printing the document to be recognized. By doing so, a recognition device based on input of a more accurate color image is realized.

Further, in the character recognition device according to the third aspect, the color information and the character recognition as to which signal of the plurality of color component signals was used as the representative value at the time of conversion by the gray conversion means. A storage means for storing the results is provided.

According to the present invention, at the time of conversion by the gray conversion means, there is provided storage means for storing color information as to which signal among a plurality of color component signals is a representative value and a character recognition result. In the post-processing, it is possible to know what color the representative value is.

Further, the character recognition apparatus according to a fourth aspect of the present invention further comprises display means for displaying the color information and the character recognition result stored in the storage means.

According to the present invention, the usability of the user is improved by displaying the color information and the character code information of the representative value on the display means.

According to a fifth aspect of the present invention, in the character recognition method for inputting a color document image and performing character recognition of the color document image, the color recognition of each pixel constituting the color document image is performed. A first step of calculating a maximum value or a minimum value of a plurality of color component signals in a signal as a representative value of the pixel; a second step of binarizing the representative value calculated in the first step; A third step of performing character recognition using the representative value binarized in the second step.

According to the present invention, the maximum value or the minimum value of a plurality of color component signals in a color signal is converted into a gray image as a representative value of the pixel by utilizing the characteristics of the color document image. After that, binarization and character recognition using the binary representative value (representative color) can be realized by a simple method.

In the character recognition method according to the present invention, the color component signal is the same as the color component signal of the ink used when printing the document to be recognized.

According to the present invention, since a color document image is often printed by a printing apparatus, a color component signal other than black and white input in the first step is printed on a document to be recognized. By using the same ink as that used at the time, a recognition method by inputting a more accurate color image is realized.

According to a seventh aspect of the present invention, there is provided a computer-readable recording medium according to the fifth or sixth aspect.
And a program for causing a computer to execute the character recognition method described in (1).

According to this invention, the program for causing a computer to execute the character recognition method according to claim 5 or 6 is recorded, whereby the character recognition operation according to claim 5 or 6 is realized by a computer. Becomes possible.

Further, in the image processing apparatus according to the present invention, in the image processing apparatus for inputting and binarizing a color image, a plurality of color component signals in a color signal of each pixel constituting the color image are provided. A representative value calculating unit that calculates a maximum value or a minimum value of the pixels as a representative value of the pixel, a color information calculating unit that calculates color information around the pixel from color components around the pixel, and the representative value calculating unit. Binarizing means for binarizing the calculated representative value, the color component signal of the pixel of the color image, and the calculated color information around the pixel.

According to the present invention, the maximum value or the minimum value of the plurality of color component signals of the color signal in each pixel of the color image is obtained, and the obtained value is set as a representative value. The color component signal of the pixel (pixel of interest),
By performing the binarization processing based on the peripheral color information (background pixel) obtained from the color component signal around the pixel, it becomes possible to cope with a change in the background density in the color image.

Further, in the image processing apparatus according to the ninth aspect, the color component signals are red, blue, and green lightness signals or cyan, magenta, and yellow lightness signals.

According to the present invention, in the image processing apparatus according to the present invention, the input color image is processed as a red, blue, and green lightness signal or a cyan, magenta, and yellow lightness signal, thereby obtaining an input image signal. Can be made to correspond to the difference in the color information.

Further, in the image processing apparatus according to the tenth aspect, the representative value is obtained from the color component signal of the pixel of the input color image used for binarization by the binarization means. This is a signal other than the color component signal used at that time.

According to the present invention, in the image processing apparatus according to the present invention, the color component signal of the pixel of the color image used for binarization is converted into a signal other than the color component signal used for obtaining the representative value. By doing so, high-precision binarization becomes possible.

Further, in the image processing apparatus according to the eleventh aspect, the color component signal for obtaining the representative value is the maximum value of the red and green color component signals.

According to the present invention, in the image processing apparatus according to the present invention, the color component signal for obtaining the representative value is the maximum value of the red and green color component signals. ), G (green), and B (black) are values close to 255 in the case of 8-bit data, whereas the values of R and G are 255 if non-background pixels are drawn in yellow. , The value of B is close to zero, and the three colors of the color when the value of R is large and the value of G and B are small, and the color where the value of G is large and the value of R and B are small are simultaneously detected. It becomes possible.

Further, in the image processing apparatus according to the twelfth aspect, the color component signal of the pixel of the input color image used for binarization by the binarization means is a blue color component signal. It is assumed that.

According to the present invention, in the image processing apparatus according to the eighth aspect, the color component signal of the pixel of the input color image used for binarization is set to the blue color component signal, thereby achieving accurate It is possible to perform a binary processing.

In the image processing apparatus according to the thirteenth aspect, the calculation of the peripheral color information is performed based on information of a pixel which has been processed before processing of the pixel.

According to the present invention, in the image processing apparatus according to the present invention, the color information (background density) around the pixel is obtained by using data processed before the pixel, thereby obtaining the accurate image data. The background density of the pixel is obtained.

In the image processing apparatus according to the present invention, a background pixel and a non-background pixel are separated from the result of the binarization processing by the binarization processing means.

According to the present invention, in the image processing apparatus according to the present invention, the background pixel and the non-background pixel are separated from the result of the binarization processing, thereby simplifying the post-processing for the object (non-background pixel). Can be realized.

In the image processing apparatus according to a fifteenth aspect, the color information calculation means calculates peripheral color information using color component information determined as a background pixel.

According to the present invention, in the image processing apparatus according to the present invention, the peripheral color information (background density) is obtained by using the color information of the pixel determined as the background pixel.
The peripheral color information not affected by the non-background pixels is obtained.

In the image processing apparatus according to the present invention, the color information calculating means may determine whether the background pixel is a background pixel or a non-background pixel if the background pixel is continuous with respect to a predetermined threshold value. The peripheral color information is calculated using the peripheral color component signals.

According to the present invention, in the image processing apparatus according to the present invention, when the background pixels are continuous with respect to a predetermined threshold value, the background color component signals are used without distinction between the background pixels and the non-background pixels. By calculating the peripheral color information, it is possible to avoid a situation in which peripheral color information cannot be obtained in an image in which non-background pixels continue continuously.

Further, in the image processing apparatus according to the seventeenth aspect, three-dimensional binarization comprising the representative value, a color component signal of the pixel of the input color image, and peripheral color information. A table is provided, and the binarization means performs a binarization process by using three feature values of the binarization table.

According to the present invention, in the image processing apparatus according to the eighth aspect, the three characteristic amounts of the representative value, the color component signal of the pixel of the input color image, and the peripheral color information are converted into three-dimensional binary values. By preparing as a conversion table, the processing at the time of the binarization processing can be performed quickly.

In the image processing apparatus according to the eighteenth aspect, the representative value is calculated after reducing the number of bits of the input color image.

According to the present invention, in the image processing apparatus according to the present invention, if the input color image has, for example, an 8-bit configuration, the input color image is converted into 5-bit data, and the representative value is calculated. By doing so, the processing can be sped up.

In the image processing apparatus according to the nineteenth aspect, a binary table is created after reducing the number of bits of an input color image.

According to the present invention, in the image processing apparatus of the seventeenth aspect, the input color image is, for example, 8
In the case of bit configuration, after converting to each 5 bits,
By creating a binarization table, the data amount can be reduced.

In the image processing apparatus according to the twentieth aspect, a specific pattern in an input color image is recognized using the processing result of the binarizing means.

According to the present invention, by performing the image processing as set forth in claims 8 to 19, separation of the background color from the object is realized, and a specific pattern such as a character can be recognized.

According to a twenty-first aspect of the present invention, in the image processing method for inputting and binarizing a color image, a plurality of color component signals in a color signal of each pixel constituting the color image are provided. A representative value calculating step of calculating a maximum value or a minimum value of the pixels as a representative value of the pixel; a color information calculating step of obtaining color information around the pixel from color components around the pixel; A binarizing step of binarizing the color component signal of the pixel of the image and the calculated color information around the pixel.

According to the present invention, the maximum value or the minimum value of the plurality of color component signals of the color signal in each pixel of the color image is obtained, and the obtained value is set as a representative value. The color component signal of the pixel (pixel of interest),
By performing the binarization processing based on the peripheral color information (background pixel) obtained from the color component signal around the pixel, it becomes possible to cope with a change in the background density in the color image.

In the image processing method according to the present invention, the color component signals are red, blue, and green lightness signals or cyan, magenta, and yellow lightness signals.

According to the present invention, in the image processing method according to the twenty-first aspect, the input color image is processed as a red, blue, and green lightness signal or a cyan, magenta, and yellow lightness signal, whereby the input image signal is processed. Can be made to correspond to the difference in the color information.

In the image processing method according to claim 23, the representative value is obtained from the color component signal of the pixel of the input color image used for binarization in the binarization step. This is a signal other than the color component signal used at that time.

According to the present invention, in the image processing method according to the twenty-first aspect, the color component signal of the pixel of the color image used for binarization is a signal other than the color component signal used for obtaining the representative value. By doing so, high-precision binarization becomes possible.

Further, in the image processing method according to claim 24, the color component signal for obtaining the representative value is the maximum value of the red and green color component signals.

According to the present invention, in the image processing method according to the twenty-first aspect, the color component signal for obtaining the representative value is the maximum value of the red and green color component signals. ), G (green) and B (black) are all 8
In the case of bit data, while the value is close to 255,
If non-background pixels are drawn in yellow, the values of R and G are 255
, The value of B is close to zero, the value of R is large,
It is possible to simultaneously detect three colors, that is, colors when the values of R and B are small, and colors where the value of G is large and the values of R and B are small.

In the image processing method according to the twenty-fifth aspect, the color component signal of the pixel of the input color image used for binarization in the binarization step is a blue color component signal. It is assumed that.

According to the present invention, in the image processing method according to the twenty-first aspect, the color component signal of the pixel of the input color image used for binarization is set to the blue color component signal, thereby achieving accurate It is possible to perform a binary processing.

In the image processing method according to the twenty-sixth aspect, the calculation of the surrounding color information is performed based on information of a pixel that has been processed before the processing of the pixel.

According to the present invention, in the image processing method according to the twenty-first aspect, the data processed before the pixel is used as the color information (background density) around the pixel to obtain an accurate image. The background density of the pixel is obtained.

Further, in the image processing method according to claim 27, a background pixel and a non-background pixel are separated from the result of the binarization processing in the binarization processing step.

According to the present invention, in the image processing method according to the twenty-first aspect, the background pixel and the non-background pixel are separated from the result of the binarization processing, thereby simplifying the post-processing for the object (non-background pixel). Can be realized.

In the image processing method according to the twenty-eighth aspect, the color information calculation step calculates surrounding color information using color component information determined as a background pixel.

According to the present invention, in the image processing method according to the twenty-first aspect, the peripheral color information (background density) is obtained by using the color information of the pixel determined to be the background pixel, so that it is not affected by non-background pixels. The surrounding color information can be obtained.

Further, in the image processing method according to claim 29, the color information calculating step is performed without discriminating between a background pixel and a non-background pixel when the background pixel is continuous with respect to a predetermined threshold value. The peripheral color information is calculated using the peripheral color component signals.

According to the present invention, in the image processing method according to the twenty-first aspect, when a large number of background pixels are continuous with respect to a predetermined threshold value, the background color component signal is used without distinction between the background pixel and the non-background pixel. By calculating the surrounding color information,
A situation in which peripheral color information cannot be obtained in an image in which non-background pixels continue continuously is avoided.

Further, in the image processing method according to claim 30, furthermore, three-dimensional binarization comprising the representative value, a color component signal of the pixel of the input color image, and peripheral color information. The method includes a step of preparing a table, and the binarization step performs a binarization process using three feature values of the binarization table.

According to the present invention, in the image processing method according to the twenty-first aspect, the three feature amounts of the representative value, the color component signal of the pixel of the input color image, and the surrounding color information are converted into three-dimensional binary values. By preparing as a conversion table, the processing at the time of the binarization processing can be performed quickly.

In the image processing method according to the thirty-first aspect, the representative value is calculated after reducing the number of bits of the input color image.

According to the present invention, when the input color image has, for example, an 8-bit configuration, it is converted into 5-bit data, and then the representative value is calculated, thereby speeding up the processing. Becomes possible.

Further, in the image processing method according to claim 32, a binary table is created after reducing the number of bits of an input color image.

According to the invention, in the image processing method according to the thirtieth aspect, the input color image is, for example, 8
In the case of bit configuration, after converting to each 5 bits,
By creating a binarization table, the data amount can be reduced.

Further, in the image processing method according to claim 33, a specific pattern in the input color image is recognized using the processing result of the binarization step.

According to the present invention, by performing the image processing as set forth in claims 8 to 19, separation of the background color from the object is realized, and a specific pattern such as a character can be recognized.

Further, in the computer-readable recording medium according to claim 34, the computer-readable recording medium according to claims 21 to 23 is provided.
3. A computer that causes a computer to execute the image processing method according to any one of 3.

According to the present invention, the above-mentioned claims 21 to 33 are provided.
21. A program for causing a computer to execute any one of the image processing methods described above.
The image processing operation described in any one of 33 can be realized by a computer.

[0077]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of a character recognition device, a character recognition method, an image processing device, an image processing method, and a computer-readable recording medium according to the present invention will be described below with reference to the accompanying drawings. Will be described. The present invention is not limited to this embodiment.

(Embodiment 1) First, a color image handled in this embodiment will be described. To digitize a color image input from a color image input device such as a color scanner, first, the color of each point is R (red), G
Decomposes into (green) and B (blue) components. This allows R,
Three monochrome images are obtained as the distribution of the shades of G and B. For each of the R channel, G channel, and B channel, A / D conversion is performed by dividing the channel into common pixels and obtaining the same number of gradations for quantization.

The digitized color images are R, G,
B is a superposition of three monochrome digital images.
If the separation into pixels in the R, G, and B channels is shared, a digital color image is generated for each pixel by the R.G.
A set of three numerical values representing the brightness of G and B (RG, B)
Is considered to have been decided. For example, when quantization is performed with 256 gradations for each channel, R.D. G and B are respectively R = 0 to 255, G = 0 to 255, and B = 0.
It takes an integer value of 255. Since 256 gradations are an 8-bit information amount, the information amount per pixel is 8 bits ×
3 = 24 bits. Representing the RGB with 8 bits each, that is, a total of 24 bits is called RGB 24-bit color expression.

Incidentally, a color document image which is generally used is rarely printed using many colors such as 100 colors or 200 colors. For example, in most cases, black and white characters have partially colored character information, or the entire page has several colored characters. The colors are cyan, magenta, and yellow (hereinafter, referred to as C, M, and Y), which are primary colors of ink used at the time of printing, and red,
Blue, green (hereinafter, referred to as R, B, G) in many cases.

Optical scanners such as scanners, facsimile machines, and digital still cameras for reading these document images are provided with R, B, and G CCDs (solid-state image sensors).
Are often used, and the output from these devices is
It is often an R, B, G color component signal.

Further, C, M, and Y, which are the primary colors of the ink used at the time of printing, and red, blue, and green output from the above-mentioned device have a complementary color relationship, and can be easily calculated by a predetermined conversion formula. It is possible. Therefore, the present invention makes it possible to recognize and reproduce a color image with a simpler device configuration by actively utilizing the features of the color document image.

Next, the configuration of the apparatus will be described. FIG. 1 is a block diagram illustrating a configuration of a character recognition device according to an embodiment of the present invention. In the figure, reference numeral 1 denotes a gray conversion circuit that creates a gray image from a color document image input from a color image input device such as a scanner, a facsimile device, or a digital still camera, and reference numeral 2 denotes a gray conversion circuit that is gray-converted by the gray conversion circuit 1. Gray image is white / black 2
A binarization circuit for converting to value data, and reference numeral 3 denotes a binarization circuit 2
Is a character recognition circuit for recognizing the character of the binary data binarized by.

FIG. 2 is a block diagram showing the configurations of the gray conversion circuit 1 and the binarization circuit 2 in FIG. The gray conversion circuit 1 comprises R (red), G (green), B (blue), C (cyan), M (magenta), Y (eralow), and K (black) constituting the input color document image. A gray image is created using the maximum value or the minimum value of a plurality of color component signals of the color signal of each pixel as a representative value of the pixel. The binarization circuit 2 binarizes the representative 8-bit image into white pixel data and black pixel data by a fixed threshold value or a discriminant analysis method.

FIG. 3 is a block diagram showing the configuration of the character recognition circuit 3 in FIG. In the figure, reference numeral 10 is 2
For example, a horizontal projection (horizontal projection in a horizontal writing document) is taken from the binary data of the value conversion circuit 2 to form a row image (character string image)
Is a character cutout unit for cutting out a character block (character element) by taking a vertical projection or tracking the connection of black pixels to the cutout line image (character string image). A feature extraction unit for extracting character size, position information, and the like from a character block (character element), code 1
Reference numeral 3 denotes a pattern memory storing standard character pattern information, and reference numeral 14 denotes a character recognition for recognizing a character from the information extracted by the feature extracting unit 12 and the information stored in the pattern memory 13 and outputting character code information. Department.

Next, the operation of the character recognition device configured as described above will be described. FIG. 4 is a flowchart illustrating a first operation procedure example according to the embodiment of the present invention. First, a color document image is input from a scanner or the like (step S11). As typical color components used for a color document image, R, G, B and C, M, Y including black (hereinafter referred to as K) in C, M, Y
Y and K are often used. Between these R, G, B and C, M, Y, "Y = 255-B" and "C = 25
5 = R "and" M = 225-G ", which can be converted by a simple conversion formula.

The conversion equation is not limited to the above-mentioned conversion equation.
In general, C = k1 × R + k2 × G + k3 × BM M = k4 × R + k5 × G + k6 × BY = k7 × R + k8 × G + k9 × B expressed. Note that k1 to k9 are coefficients.

In general, R, G, B or C, M, Y,
(K) is used when a CCD used in an optical reading device such as a scanner, a facsimile device, or a digital still camera outputs R, G, and B color component signals, and is used in a printer or the like. This is because the primary colors of the ink used in the printing apparatus are C, M, Y, and (K).

On the other hand, a color document image which is a target document of the character recognition device is often printed by some printing device. Therefore, color parts other than black and white are R, G, B
Or, it is often represented by C, M, Y. Also,
The background (paper) portion of a document is often white or high in brightness, and the brightness value of a character portion is generally low in order to distinguish it from the background.

After executing step S11 in FIG. 4, the minimum (or maximum) value of R, G, B (C, M, Y, (K)) is calculated as a representative value (step S1).
2). Here, an example of creating a representative value 8-bit image from R, G, and B 8-bit 256 gradations (total 24 bits) will be described.

For example, in the R, G, B color signals, a red (R) character has an R, G, B value (255, 0, 0), and a green (G) character, R, G, B, The G and B values are (0, 2
55, 0), the R, G, B values of blue (B) are (0, 0, 255). In the case of a black character, it is (0, 0, 0), and the background (paper) part is (25, 0).
5, 255, 255). FIG. 5 shows a summary of the above.

As shown in FIG. 5, in the character portion, R,
By utilizing the fact that one of the G and B values becomes smaller (becomes 0 in this example) and using the minimum value of the R, G and B values as a representative value, the character portion and the background are independent of the color. It is possible to separate the parts.

Next, when the representative value is calculated in step S12 of FIG. 4, subsequently, a binarization process is executed (step S13). That is, at the stage of the representative value, in the case of FIG. 5, each of R, G, and B is 8 bits (256 gradations).
Are only 8 bits per pixel, and are not binary. As a method of binarizing a representative value 8-bit image, binarization may be performed with a fixed threshold value, or a known method used in various places such as a discriminant analysis method may be adopted.

Subsequently, the binary data binarized in step S13 is input to the character recognition circuit 3, and the character recognition processing is executed (step S14). That is, a character string is cut out by the line cutout unit 10. Furthermore, a character block is cut out from the line image (character string image) cut out by the character cutout unit 11 by taking a vertical projection or tracing the connection of black pixels, and the size and position of the character are extracted by the feature extraction unit 12. I do. Thereafter, the character recognition unit 14 performs character recognition by referring to the dictionary of the pattern memory 13, and outputs character code information.

Next, the character recognition result (character code information) described above and the color information of the color of the character based on which of the R, G, and B values was used as the representative value are also combined. It is stored in a memory provided in the apparatus or an external memory, or displayed on a display or the like (step S
15).

FIG. 7 is a block diagram showing the structure of another character recognition device. In the structure of the character recognition device shown in FIG. 1, character code information and color information are stored at the subsequent stage of the character recognition circuit 3. Memory 4 is provided. FIG. 8 is a flowchart showing an operation procedure of the character recognition device of FIG. 7, and is a flowchart showing a second operation procedure example. In this operation procedure, steps S23 and S26 are added to the flowchart of FIG.
Is added.

That is, in step S22, R, G, B
After calculating the minimum value (or maximum value) of (C, M, Y, (K)) as a representative value, R, G, B (C, M, Y,
Which value of (K)) was used as the representative value is stored in the memory 4 (step S23). In step S26, the character code information of the character recognized in step S25 is stored in the memory 4. The other processes are the same as those in FIG.

As described above, by storing which of the R, G, and B values is used as the representative value, it is possible to obtain information on what color of the character was written together with the character recognition result. it can.

By the way, instead of using the method according to the above-described embodiment, only the G component is extracted as a conversion method from “color to gray” which is often used in the related art.
In the case of a green character, it is impossible to distinguish it from the background.

In this embodiment, the description has been made using R, G, and B. However, the same applies to C, M, Y, and (K). Also, in this embodiment,
Although the R, G, and B values of the black character are set to (0, 0, 0), the black character may be set to (255, 255, 255) depending on the image format.
In some cases, the maximum value of the R, G, and B values is used as the representative value.

(Embodiment 2) FIG. 9 is a block diagram showing a configuration of an image processing apparatus according to Embodiment 2 of the present invention. This image processing apparatus includes a color image input unit 100 for inputting R, G, and B color image signals, and a gamma conversion for converting each 8-bit image signal of the R, G, and B color image signals into each 5-bit signal. Unit 101, a maximum value calculation unit 102 for obtaining the maximum value of the R and G image signals, and a background density calculation unit 103 for obtaining the background density from the R, G and B color image signals
And a binarization processing unit 104 for binarizing the maximum value and the background density, and a binarization table 105 in which three-dimensional features are tabulated.

Next, an image processing apparatus configured as described above will be described. First, the color image input unit 100
Inputs an R, G, B color image signal from a color input device such as a digital copying machine or a color image scanner, or a computer. The gamma converter 101 converts an 8-bit image into a 5-bit image for each of the R, G, and B color image signals. Further, the maximum value calculation unit 102 obtains the maximum value of the values of R and G and sets the maximum value as a representative value.

As will be described later, the background density calculating section 103 refers to the preceding 32 pixels from the 5 bits of the R, G, B color image signal to determine the background density. The background density, the representative value, and the pixel value of B not used for calculating the representative value are sent to the binarization processing unit 104. Binarization processing unit 104
Performs a binarization process on the representative value MAX (R, G) 5 bits, the background density (5 bits), and B (5 bits) using the binarization table 105. Note that the binarization table 105 is
It has a three-dimensional feature amount as described later.

Here, the color image signal is represented by R,
G, B, C, M, Y or C, M, Y,
It may be a value such as K. Also, the maximum value calculation unit 102
Although R and G are used here when calculating the representative value, G and B or R and B may be used, or any combination of C, M, Y, and K may be used. .
Further, instead of comparing these two signals, a representative value may be obtained by comparing three or more signals.

Next, as color information around the pixel,
The operation for obtaining the background density will be described. In the case of a digital color copying machine, a digital color scanner, or the like, data is read line by line in the main scanning direction. If the read data is not processed sequentially, it is necessary to store the data of the previous line in a line memory. Therefore, color information (= background density) around the pixel is calculated using only data within one line, and only data input before the pixel of interest. The calculation of the background density will be described in detail below.

FIG. 10 is an explanatory diagram showing a method of calculating the background density according to the second embodiment of the present invention. FIG. 10 shows an example in which the background density is calculated by looking at the values of 32 pixels in front of the target pixel 51. Here, as the peripheral color information (= background density), “the difference between the maximum value of R and G and the value of B” is one feature amount.

That is, in FIG. 10A, the target pixel 51
In the 32 pixels in front of the pixel determined to be the background pixel 52 by the binarization process, let (intens (= Max (R, B) -B)) be the background density.

On the other hand, as shown in the image pattern shown in FIG.
When all eight pixels are the non-background pixels 53a, the (in) of all 32 pixels in front of the pixel of interest 51 regardless of the binarization result
(sum of tens / 32) is set as the background density.

Therefore, it is possible to simultaneously detect non-background pixels (objects) of three colors written on a white background by using the feature amount “difference between the maximum value of R and G and the value of B”. Become. That is, in the case of a white background, all values of R, G, and B are close to 255 in the case of 8-bit data, whereas the values of R and G are 255 when a non-background pixel is drawn in yellow. And the value of B is close to zero. Similarly, R
Can be detected at the same time when the value of G is large and the values of G and B are small, and when the value of G is large and the values of R and B are small.

In the above-described processing, the background pixel 5
By obtaining the surrounding color information (= background density) using only the color information of the pixel determined to be 2, it is possible to obtain the surrounding color information without being affected by the non-background pixels.

However, when non-background pixels continue continuously (FIG. 10B), there arises a problem that peripheral color information cannot be obtained. Therefore, in consideration of the size of the object to be extracted, for example, if the maximum number of pixels of the extracted object is 16 pixels, the object cannot continue more than 16 pixels. If the non-background pixels continue, the background density is calculated from all the pixel values immediately before the target pixel 51.

The method of calculating the background density is shown in the flowchart of FIG. First, each parameter is initialized (step S31). That is, mean1 (the front 32 pixels of the pixel of interest), mean (of the front 32 pixels of the pixel of interest, the pixel determined to be the background pixel), and flag are initialized to 0. Subsequently, a binarization process is executed (step S3).
2).

That is, rst (x) = binTable (max (x), bl
ue (x), mean) rst (x) = 0 White pixel (background) rst (x) = 1 Black pixel (non-background) Subsequently, the background density is obtained according to the illustrated algorithm (step S33).

In FIG. 11, max (x) = Max (red (x), green
(X)) intens (x) = Max (red (x), green
n (x))-blue (x) mean1 = Σintens (x), mean = Σin
tens (x).

The character recognition method and image processing method described above can be programmed, recorded on a computer-readable recording medium, and executed on a computer. Further, a part of the character recognition method and the image processing method may be provided on a network, and may be realized through a communication line.

That is, according to the character recognition method and the image processing method described in the present embodiment, as shown in FIG. 12, a program prepared in advance is transferred to a computer (CPU 2) such as a personal computer or a workstation.
0). This program is operated by operating the keyboard 25 or the like, and the memory 21, the hard disk 24, the floppy (registered trademark) disk 2
7. The program is recorded on a computer-readable recording medium such as a CD-ROM 26, an MO, and a DVD, and is executed by being read from the recording medium by a computer (CPU 20). If necessary, the character recognition or image processing data can be transmitted and received from the communication device 22 to an external device.

As shown in FIG. 13, this program can be distributed to devices 31 to 33 such as personal computers via the recording medium via a network such as the Internet 30.

[0118]

As described above, according to the character recognition device of the present invention (claim 1), the color document image which is the target document of the character recognition device has R, G,
By utilizing the fact that one of the B values becomes small (becomes 0 in this example) and using the minimum value of the R, G, and B values as a representative value, the character portion and the background portion can be separated regardless of the color. After converting the maximum value or the minimum value of a plurality of color component signals in a color signal into a gray image as a representative value of the pixel by using a feature of a color document image that the color document image can be divided. , Binarization, part 2
Since the character is recognized using the representative value (representative color) of the value, the character of the color document image can be recognized with a device having a simpler configuration.

According to the character recognition device of the present invention (claim 2), since a color document image is often printed by some printing device, a color component signal other than black and white is recognized as a recognition target. By using the same ink as that used when printing the original, the representative value under the same conditions as when printing is used, so that more accurate character recognition of a color image can be performed.

Further, according to the character recognition device of the present invention, the color information and the character recognition as to which signal of the plurality of color component signals was used as the representative value at the time of conversion by the gray conversion means. By providing storage means for storing the result, in post-processing such as when using the recognition result,
It is possible to quickly confirm what color the representative value is.

According to the character recognition device of the present invention, the display means displays the color information and the character code information of the representative value, so that the character code information and the color information of the recognition result are displayed. The user can immediately confirm the information on a display unit such as a display.

According to the character recognition method of the present invention (claim 5), similarly to claim 1, the maximum value or the minimum value of a plurality of color component signals in a color signal is utilized by utilizing the characteristics of a color document image. After the value is converted into a gray image as a representative value of the pixel, the image is binarized, and the character is recognized using the binary representative value (representative color). Therefore, the character of the color document image is recognized by a simpler method. be able to.

According to the character recognition method of the present invention (Claim 6), since a color document image is often printed by a certain printing device, a color document image other than black and white input in the first step is used. By making the color component signal the same as the ink used when printing the document to be recognized, a representative value under the same conditions as at the time of printing is used, so that more accurate character recognition of a color image can be performed. it can.

According to the computer-readable recording medium of the present invention (claim 7), a program for causing a computer to execute the character recognition method according to claim 5 or 6 is recorded. Alternatively, the character recognition operation described in 6 can be realized by a computer.

Further, according to the image processing apparatus of the present invention, the maximum value or the minimum value of a plurality of color component signals of the color signal in each pixel of the color image is obtained, and the obtained value is set as a representative value. The binarization process is performed based on the representative value, the color component signal of the pixel (target pixel) of the input color image, and the peripheral color information (background pixel) obtained from the color component signal of the periphery of the pixel. Therefore, even if the background density of the input color image changes, binarized data of the object image can be obtained with high accuracy.

According to the image processing apparatus of the present invention, the input color image is converted to a red, blue, and green lightness signal or cyan, magenta, and yellow. Since the brightness signal is processed as described above, it is possible to cope with a difference in color information of the input image signal.

Further, according to the image processing apparatus of the present invention (claim 10), in the image processing apparatus of claim 8,
Since the color component signal of the pixel of the color image used for binarization is a signal other than the color component signal used for obtaining the representative value, high-precision binarization becomes possible.

Further, according to the image processing apparatus of the present invention (claim 11), in the image processing apparatus of claim 8,
Since the color component signal for obtaining the representative value is the maximum value of the red and green color component signals, non-background pixels (objects) of three colors written on a white background can be simultaneously detected.

Further, according to the image processing apparatus of the present invention (claim 12), in the image processing apparatus of claim 8,
By making the color component signal of the pixel of the input color image used for binarization a blue color component signal, accurate binarization processing becomes possible.

Further, according to the image processing apparatus of the present invention (claim 13), in the image processing apparatus of claim 8,
Since the color information (background density) around the pixel is determined using data processed before the pixel, the background density of the pixel can be accurately obtained.

Further, according to the image processing apparatus of the present invention (claim 14), in the image processing apparatus of claim 8,
Since the background pixel and the non-background pixel are separated from the result of the binarization process, post-processing on the object (non-background pixel) can be easily performed.

Further, according to the image processing apparatus of the present invention (claim 15), in the image processing apparatus of claim 8,
Since peripheral color information (background density) is obtained using color information of a pixel determined as a background pixel, peripheral color information not affected by non-background pixels can be obtained.

According to the image processing apparatus of the present invention (claim 16), in the image processing apparatus of claim 8,
If the background pixels are continuous with respect to a predetermined threshold value, the peripheral color information is calculated using the peripheral color component signals without distinction between the background pixels and the non-background pixels. In this case, a situation in which peripheral color information cannot be obtained is avoided, and a correct background density can be obtained.

According to the image processing apparatus of the present invention (claim 17),
By preparing the three characteristic amounts of the representative value, the color component signal of the pixel of the input color image, and the surrounding color information as a three-dimensional binarization table, the processing at the time of the binarization processing is quick. , The processing efficiency is improved.

Further, according to the image processing apparatus of the present invention (claim 18), in the image processing apparatus of claim 8,
If the input color image has, for example, an 8-bit configuration, it is converted into 5-bit data and then the representative value is calculated, so that the representative value can be calculated efficiently.

Further, according to the image processing apparatus of the present invention (claim 19), in the image processing apparatus of claim 17, when the input color image has, for example, an 8-bit configuration, each of the input color images has 5 bits. After the conversion into bits, a binary table is created, so that the data amount can be reduced, and the effective use of the memory and the reduction of the capacity are realized.

Further, according to the image processing apparatus of the present invention (claim 20), by performing the image processing of claims 8 to 19, the background color and the object can be reliably separated. , Characters and other specific patterns can be accurately recognized.

According to the image processing method of the present invention (claim 21), the maximum value or the minimum value of a plurality of color component signals of a color signal in each pixel of a color image is obtained, and the value is set as a representative value. The binarization process is performed based on the representative value, the color component signal of the pixel (target pixel) of the input color image, and the peripheral color information (background pixel) obtained from the color component signal of the periphery of the pixel. Therefore, even if the background density of the input color image changes, binarized data of the object image can be obtained with high accuracy.

Further, according to the image processing method of the present invention (claim 22), in the image processing method of claim 21, an input color image is converted into red, blue, and green lightness signals or cyan, magenta, and yellow. Since it is processed as the brightness signal of the input image signal, it is possible to cope with the difference in the type of the color information of the input image signal.

According to the image processing method of the present invention (claim 23), in the image processing method of claim 21, the color component signal of the pixel of the color image used for binarization is represented by a representative value. Is obtained as a signal other than the color component signal used in the calculation of .times..times.

According to the image processing method of the present invention (claim 24), in the image processing method of claim 21, the color component signal for obtaining the representative value is the maximum value of the red and green color component signals. Therefore, non-background pixels (objects) of three colors written on a white background can be simultaneously detected.

According to the image processing method of the present invention (claim 25), in the image processing method of claim 21, the color component signal of the pixel of the input color image used for binarization is obtained. , And a blue color component signal, accurate binarization processing can be performed.

According to the image processing method of the present invention (claim 26), in the image processing method of claim 21, color data around the pixel is determined by using data processed before the pixel. (Background density)
The background density of the pixel can be accurately obtained.

According to the image processing method of the present invention, the background pixel and the non-background pixel are separated from the result of the binarization processing. Object in image (non-background pixel)
Post-processing can be easily performed.

Further, according to the image processing method of the present invention (claim 28), in the image processing method of claim 21, the peripheral color information (background density) is obtained by using the color information of the pixel determined as the background pixel. ), It is possible to obtain peripheral color information that is not affected by non-background pixels.

Further, according to the image processing method of the present invention (claim 29), in the image processing method of claim 21, when a large number of background pixels continue with respect to a predetermined threshold value,
Since the peripheral color information is calculated using the peripheral color component signals without distinction between the background pixel and the non-background pixel, a situation in which the peripheral color information cannot be obtained in an image in which the non-background pixels continue continuously is avoided, Correct peripheral color information can be obtained.

According to the image processing method of the present invention (claim 30), in the image processing method of claim 21, the representative value, the color component signal of the pixel of the input color image, and the peripheral color information , Are prepared as a three-dimensional binarization table and used at the time of binarization, so that the processing at the time of binarization can be performed efficiently.

Further, according to the image processing method of the present invention (claim 31), when the input color image has, for example, an 8-bit configuration, converts the input color image into 5-bit data, By performing the calculation processing, the processing is performed with a small amount of data, so that the processing efficiency of the representative value calculation is improved.

According to the image processing method of the present invention (claim 32), in the image processing method of claim 30, when the input color image has, for example, an 8-bit configuration, each of the input color images has a 5-bit configuration. Since the binarization table is created after the conversion into bits, the amount of data can be reduced, and the memory capacity can be effectively used and the size can be reduced.

According to the image processing method of the present invention (claim 33), by performing the image processing according to claims 21 to 32, the separation between the background color and the object becomes more reliable. , Characters and other specific patterns can be more accurately recognized.

According to the computer-readable recording medium of the present invention (claim 34), claim 21
By recording a program for causing a computer to execute any one of the image processing methods according to any one of (1) to (33), the image processing operation according to any one of claims 21 to 33 can be realized by the computer.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a character recognition device according to a first embodiment of the present invention.

FIG. 2 is a block diagram illustrating a configuration of a gray conversion circuit and a binarization circuit in FIG. 1;

FIG. 3 is a block diagram illustrating a configuration of a character recognition circuit in FIG. 1;

FIG. 4 is a flowchart illustrating a first operation procedure example according to the first embodiment of the present invention;

FIG. 5 is an explanatory diagram illustrating an example of determining a representative value from RGB values according to the first embodiment of the present invention.

FIG. 6 is an explanatory diagram showing a case where only a G component is extracted from RGB values.

FIG. 7 is a block diagram showing a configuration of another character recognition device according to the first embodiment of the present invention.

FIG. 8 is a flowchart illustrating a second operation procedure example according to the first embodiment of the present invention;

FIG. 9 is a block diagram illustrating a configuration of an image processing apparatus according to a second embodiment of the present invention;

FIG. 10 is an explanatory diagram illustrating an example of calculating a background density according to the second embodiment of the present invention;

FIG. 11 is a flowchart illustrating a calculation process of a background density according to the second embodiment of the present invention;

FIG. 12 is a block diagram illustrating an example of a computer system that implements a character recognition method and an image processing method according to the present invention by software.

FIG. 13 is a block diagram showing an example of a system for realizing a character recognition method and an image processing method according to the present invention on a network.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Gray conversion circuit 2 Binarization circuit 3 Character recognition circuit 4 Memory 10 Line extraction part 11 Character extraction part 12 Feature extraction part 14 Character recognition part 100 Color image input part 101 Gamma conversion part 102 Maximum value calculation part 103 Background density calculation part 104 binarization processing unit 105 binarization table

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) H04N 1/46 H04N 1/40 103C 1/46 Z

Claims (34)

[Claims] 1. A character recognition apparatus for inputting a color document image and performing character recognition on the color document image, comprising: a maximum value or a minimum value of a plurality of color component signals in a color signal of each pixel constituting the color document image. Conversion means for converting a gray image converted by the gray conversion means into a gray image as a representative value of the pixel, binarization means, and character recognition from the binary data of the binarization means And a character recognizing means for performing the following. 2. The character recognition device according to claim 1, wherein the color component signal is the same as a color component signal of ink used when printing a document to be recognized. 3. The image processing apparatus according to claim 1, further comprising a storage unit for storing color information and a character recognition result indicating which signal of the plurality of color component signals is a representative value at the time of conversion by the gray conversion unit. The character recognition device according to claim 1 or 2, wherein: 4. The character recognition apparatus according to claim 3, further comprising display means for displaying the color information and the character recognition result stored in said storage means. 5. A character recognition method for inputting a color document image and performing character recognition on the color document image, comprising: a maximum value or a minimum value of a plurality of color component signals in a color signal of each pixel constituting the color document image. As a representative value of the pixel, a second step of binarizing the representative value calculated in the first step, and a representative value binarized in the second step. And a third step of performing character recognition by using the character recognition method. 6. The character recognition method according to claim 5, wherein the color component signal is the same as a color component signal of ink used when printing a document to be recognized. 7. A computer-readable recording medium on which a program for causing a computer to execute the character recognition method according to claim 5 or 6 is recorded. 8. An image processing apparatus for inputting and binarizing a color image, wherein a maximum value or a minimum value of a plurality of color component signals in a color signal of each pixel forming the color image is set as a representative value of the pixel. A representative value calculating means for calculating; a color information calculating means for obtaining color information around the pixel from color components around the pixel; and a representative value calculated by the representative value calculating means and a value of the pixel in the color image. An image processing apparatus comprising: a binarizing unit that binarizes a color component signal and calculated color information around the pixel. 9. The image processing apparatus according to claim 8, wherein the color component signals are red, blue, and green lightness signals or cyan, magenta, and yellow lightness signals. 10. A color component signal of a pixel of an input color image used for binarization by said binarization means, other than a color component signal used for obtaining said representative value. The image processing apparatus according to claim 8, wherein: 11. The image processing apparatus according to claim 8, wherein the color component signal for obtaining the representative value is a maximum value of red and green color component signals. 12. The color component signal according to claim 8, wherein the color component signal of the pixel of the input color image used for binarization by the binarization means is a blue color component signal. Image processing device. 13. The image processing apparatus according to claim 8, wherein the calculation of the surrounding color information is performed based on information of a pixel that has been processed before the processing of the pixel. 14. The image processing apparatus according to claim 8, wherein a background pixel and a non-background pixel are separated from a result of the binarization processing by the binarization processing unit. 15. The image processing apparatus according to claim 8, wherein the color information calculation means calculates surrounding color information using color component information determined as a background pixel. 16. The color information calculating means, when the number of the background pixels is larger than a predetermined threshold value, determines the peripheral color information using the peripheral color component signal without distinction between the background pixel and the non-background pixel. The image processing apparatus according to claim 8, wherein the calculation is performed. 17. A three-dimensional binarization table comprising the representative value, a color component signal of the pixel of the input color image, and peripheral color information, wherein the binarization means comprises 9. The image processing apparatus according to claim 8, wherein the binarization process is performed using three feature amounts of the conversion table. 18. The image processing apparatus according to claim 8, wherein the representative value is calculated after reducing the number of bits of the input color image. 19. The image processing apparatus according to claim 17, wherein the binarization table is created after reducing the number of bits of the input color image. 20. The image processing apparatus according to claim 8, wherein a specific pattern in the input color image is recognized using a processing result of the binarizing unit. 21. An image processing method for inputting and binarizing a color image, wherein a maximum value or a minimum value of a plurality of color component signals in a color signal of each pixel constituting the color image is set as a representative value of the pixel. A representative value calculating step of calculating; a color information calculating step of obtaining color information around the pixel from color components around the pixel; a color component signal of the pixel of the representative value and the color image;
And a binarizing step of binarizing the calculated color information around the pixel. 22. The image processing method according to claim 21, wherein the color component signals are red, blue, and green lightness signals or cyan, magenta, and yellow lightness signals. 23. A color component signal of a pixel of an input color image used for binarization in the binarization step, other than a color component signal used for obtaining the representative value. The image processing method according to claim 21, wherein: 24. The image processing method according to claim 21, wherein the color component signal for obtaining the representative value is a maximum value of red and green color component signals. 25. The color image processing apparatus according to claim 21, wherein the color component signal of the pixel of the input color image used for binarization in the binarization step is a blue color component signal. Image processing method. 26. The image processing method according to claim 21, wherein the calculation of the peripheral color information is performed based on information of a pixel that has been processed before the processing of the pixel. 27. The image processing method according to claim 21, wherein a background pixel and a non-background pixel are separated from a result of the binarization processing in the binarization processing step. 28. The image processing method according to claim 21, wherein said color information calculating step calculates peripheral color information using color component information determined as a background pixel. 29. The color information calculating step, when the number of the background pixels is continuous with respect to a predetermined threshold value, determines the surrounding color information using the surrounding color component signals without distinction between the background pixels and the non-background pixels. The image processing method according to claim 21, wherein the calculation is performed. 30. A step of preparing a three-dimensional binarization table including the representative value, a color component signal of the pixel of the input color image, and peripheral color information. 22. Performs a binarization process using three feature amounts of the binarization table.
The image processing method according to 1. 31. The image processing method according to claim 21, wherein the representative value is calculated after reducing the number of bits of the input color image. 32. The image processing method according to claim 30, wherein the binarization table is created after reducing the number of bits of the input color image. 33. The image processing method according to claim 21, wherein a specific pattern in the input color image is recognized using a processing result of the binarization step. 34. A computer-readable recording medium on which a program for causing a computer to execute the image processing method according to claim 21 is recorded.