JP7362405B2 - Image processing device, image processing method, and program - Google Patents

Description

The present invention relates to an image processing device, an image processing method, and a program, and more particularly to a technique for performing different types of image compression for text areas and other areas such as background areas.

This type of technology divides the image to be compressed into a text area and other photo areas and background areas, compressing the text area using the MMR method, and compressing the photo area and background area using the JPEG method. . This prevents the occurrence of image deterioration called mosquito noise that may occur when the text portion of an image is compressed using the JPEG method.

Patent Document 1 solves the problem that when image regions are divided in this way and the compression method is different for each region, the color information changes when extracting the representative color of the text region from the compressed image. Disclosed. In order to solve this problem, Patent Document 1 states that compression is performed in two stages, representative colors are extracted for images with a low compression ratio, and changes in the color information are reduced.

Japanese Patent Application Publication No. 2013-125994

However, in Patent Document 1, even if an attempt is made to reduce the deterioration of character color information by reducing the compression rate through stepwise compression, even if the color change is small, it will occur in accordance with the reduced compression rate. Become. This may cause image quality deterioration in the final image, although the degree may vary.

An object of the present invention is to provide a technique that can prevent text color changes in a finally generated image even if text color changes occur due to text color extraction in a compressed image.

In order to achieve the above object, an image processing apparatus according to one aspect of the present invention includes a reduction means for converting a first image into a second image having a lower resolution than the first image, and a second image obtained by the reduction means. an extraction means for extracting a character color in a character region of an image; and a relationship between a color value of the character color extracted by the extraction means and a color value of a background color in the character region of the second image; An image including a correction means for correcting based on the number of pixels forming an edge of a character in the character area and the number of pixels forming the character, and the character area whose character color has been corrected by the correction means, The image forming apparatus is characterized by comprising a generating means for generating a compressed image in which the first image is compressed.

According to the present invention, even if text color changes occur due to text color extraction in a compressed image, it is possible to prevent text color changes in the finally generated image.

FIG. 3 is a schematic diagram illustrating that a character color changes due to character extraction in a compressed image according to an embodiment. FIG. 2 is a schematic diagram showing a pixel in FIG. 1 in an enlarged manner. 1 is a block diagram showing the configuration of an image processing system according to an embodiment of the present invention. 4 is a block diagram showing the configuration of the MFP shown in FIG. 3. FIG. FIG. 2 is a block diagram showing the configuration of an image processing section realized by a data processing section according to an embodiment of the present invention. (a) to (e) are diagrams illustrating images generated by image processing by the data processing unit according to the first embodiment of the present invention. FIG. 3 is a schematic diagram for explaining processing of a character color extraction unit according to the embodiment. (a) and (b) are schematic diagrams for explaining details of character color extraction processing according to the embodiment. FIG. 3 is a diagram schematically showing a window for edge detection used in an embodiment of the present invention. FIG. 3 is a schematic diagram for specifically explaining a correction coefficient according to an embodiment of the present invention. FIG. 3 is a diagram showing the results of correcting the representative colors of characters by the correction according to the first embodiment of the present invention. (a), (b), and (c) are flowcharts showing processing executed by the data processing unit according to the first embodiment of the present invention. 3 is a flowchart showing text color correction processing according to the first embodiment of the present invention. FIG. 7 is a schematic diagram showing an example of an image in which the background color and size change for each character according to the second embodiment of the present invention. FIG. 7 is a schematic diagram showing the result of character color correction for each character according to the second embodiment of the present invention. It is a flowchart which shows the process performed by the data processing part based on 2nd Embodiment of this invention.

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments for carrying out the present invention will be described in detail with reference to the drawings. However, the components described in this embodiment are merely examples, and the scope of the present invention is not intended to be limited thereto. Further, not all combinations of the constituent elements described in the embodiments are essential to the means for solving the problem, and various modifications and changes are possible.

First, before describing embodiments of the present invention, it will be explained that character color changes occur when character color is extracted from a compressed image. As an example, it will be explained that by image compression, the colors of 2×2 pixels in the vertical and horizontal directions in an input image with a resolution of 300 dpi are reduced to the color of 1 pixel in an image with a resolution of 150 dpi. FIG. 1 is a schematic diagram for explaining this color change. In FIG. 1, reference numerals 601, 602, and 603 indicate input images (first images) with a resolution of 300 dpi, and represent characters on a white background, characters on a colored background, and inverted characters, respectively. Reference numerals 604, 605, and 606 indicate reduced images (second images) with a resolution of 150 dpi obtained by compressing the data of the input image. Reference numerals 607, 608, and 609 indicate binary images representing text colors obtained by extracting text colors from the reduced images 604 to 606.

A reduced image 604 obtained by compressing the input image 601 shows that the edges of the characters have become thinner due to the reduction (gray (R, G, B) = (128, 128, 128)). This is because, as shown in FIG. 2, the colors of 2×2 pixels in the vertical and horizontal directions in the input image 601 are combined (averaged) into the color of one pixel. Note that FIG. 2 is an enlarged view showing a pixel 610 in the input image 601 and a pixel 611 in the reduced image 604. When text color extraction is performed on the reduced image 604 in which the color of one pixel is an averaged color as described above, the text color in the resulting binary image 607 is compared with the text in the input image 601. The color becomes brighter (lighter in density). That is, since the average is calculated from all black pixels and all gray pixels of the character in the reduced image, the character color becomes brighter ((R, G, B) = (80, 80, 80)).

Similarly, for the characters in the input image 602, color changes occur due to character color extraction from the compressed image. In other words, for black characters (black (R, G, B) = (0, 0, 0)) on a pink background in the input image 602, the edges of the characters become lighter in the reduced image 605, and the background color becomes pink. becomes a mixed state ((R, G, B)=(128, 100, 128)). Then, when text color extraction is performed on this reduced image 605, the text color of the binary image 608 becomes a relatively bright color, and the background color is pink (pink (R, G, B) = (255, 200 , 255)) is mixed ((R, G, B) = (100, 80, 100)). The same applies to the input image 603 of inverted characters, and the white characters on a black background (white (R, G, B) = (255, 255, 255)) have dark edges in the reduced image 606. (Gray (R, G, B) = (128, 128, 128)). Then, a binary image 609 obtained by performing text color extraction on the reduced image 606 has a dark text color ((R, G, B)=(140, 140, 140)). This is to calculate the average from all white pixels and all gray pixels in the reduced image 606.

As described above, when text color extraction is performed on a compressed reduced image, a problem may arise in that the text color of the finally obtained image changes.

In an embodiment of the present invention, the extracted text color is corrected depending on the degree to which the text color is influenced by the background color due to image reduction. This makes it possible to reduce changes in the finally obtained text color from the text color in the original image.

(First embodiment)
FIG. 3 is a block diagram showing the configuration of an image processing system according to an embodiment of the present invention. The image processing system of this embodiment is configured by connecting a multifunction peripheral (MFP) 101 and a client PC 102 via a network 103. In FIG. 3, dashed lines 104 and 105 each indicate the flow of processing, and dashed line 104 indicates processing in which the user uses the scanner of the MFP 101 to read a paper document. At that time, the user can use the operation unit (203 in FIG. 4) of the MFP 101, which will be described later, to set the destination to which the scanned image is to be sent (for example, the client PC 102) and various settings related to scanning and transmission. As settings, the user can specify the color mode, file format (for example, JPEG, TIFF, PDF, PDF (high compression)), etc. Below, we will explain the case where PDF (high compression) is specified as the data format. The details of PDF (high compression) will be described later.A broken line 105 indicates that data is generated using the software or hardware functions of the MFP 101 based on various specified settings, and is sent to the specified destination. The image is transmitted to the client PC 102 in a file format such as PDF, so it can be viewed using a general-purpose viewer included in the client PC 102.

FIG. 4 is a block diagram showing the configuration of MFP 101 shown in FIG. 3. As shown in FIG. 4, the MFP 101 of this embodiment includes a scanner unit 201 that is an image input device, a printer unit 202 that is an image output device, a control unit 204 that controls the entire MFP, an operation unit 203 that is a user interface, etc. It is configured with The control unit 204 is a controller that inputs and outputs image information and device information by making signal connections with the scanner section 201, printer section 202, and operation section 203, and on the other hand, by making a signal connection with the LAN 209. A CPU 205 constituting the control unit 204 is a processor that controls the entire system of the MFP 101. Similarly, the RAM 206 is a system work memory for the operation of the CPU 205, and is also an image memory for temporarily storing image data. Further, the ROM 210 is a boot ROM, and stores programs such as a system boot program. Furthermore, the storage unit 211 is a nonvolatile storage medium such as a hard disk drive, and stores system control software and image data. The operation unit I/F 207 is an interface unit with the operation unit (UI) 203 and outputs image data to be displayed on the operation unit 203 to the operation unit 203. Further, the operation unit I/F 207 serves to transmit information instructed by the user of the image processing apparatus via the operation unit 203 to the CPU 205. A Network I/F 208 connects this image processing apparatus to a LAN 209 and inputs and outputs data. For example, compressed data in PDF format is transmitted to another device, or compressed data in PDF format is received from another device. The above devices are arranged on the system bus 216. System bus 216 connects devices within control unit 204 to communicate information. Further, the ImageBus I/F 212 is a bus bridge that connects the system bus 216 and an image bus 217 that transfers image data at high speed, and converts the data structure. The image bus 217 is composed of, for example, a PCI bus or IEEE1394. The following devices are arranged on the image bus 217. The RIP unit 213 analyzes a PDL (page description language) code and develops it into a bitmap image of a specified resolution, which is what is called rendering processing. The device I/F 214 is connected to the scanner section 201, which is an image input device, via a signal line 218, and to the printer section 202, which is an image output device, via a signal line 219. Performs asynchronous conversion. The data processing unit 215 generates a PDF (highly compressed) by performing processing such as region determination, compression processing, and PDF file generation. The generated PDF (highly compressed) is sent to a specified destination (for example, client PC 102) via Network I/F 208 and LAN 209. The data processing unit 215 can also decompress compressed data received via the Network I/F 208 and the LAN 209. The expanded image will be sent to the printer unit 202 via the device I/F 214 and printed.

The data processing unit 215 shown in FIG. 4 constitutes an image processing unit that generates a PDF (highly compressed) from input data, as outlined above. FIG. 5 is a block diagram showing the configuration of the image processing section realized by the data processing section 215, in which each section is used to generate output data (PDF (highly compressed)) from input data (RGB multivalued image data). The processing section is shown. The data processing unit 215 may be configured so that the processor executes a computer program to function as each processing unit shown in FIG. It may be configured.

In FIG. 5, a gray conversion unit 301 generates gray multi-value image data based on input data (RGB multi-value image data) read by the scanner unit 201. FIG. 6A shows an example of input data (RGB multi-value image data) read by the scanner unit 201. The input data includes a character string 501 of "EF", a character string 502 of "EF" on a colored background, a character string 503 of "EF" in outline characters, and a photographic image 504. Based on input data including these, gray image data with a luminance Y defined by the following equation (1) is generated.

Y=0.299×R+0.587×G+0.114×B...(1)
In this embodiment, the luminance Y defined by the above equation (1) in the general YUV color space is used as the gray signal, but the present invention is not limited thereto.

The binarization unit 302 generates binary image data based on the gray multi-value image data obtained by the gray conversion unit 301. The binarization method used in this embodiment is a method in which a single threshold value is calculated from a histogram obtained from gray multivalued image data, and binarization is performed using the threshold value. FIG. 6(b) shows an example of a binary image generated by the binarization unit 302 based on the gray multivalued image shown in FIG. 6(a).

Referring again to FIG. 5, the first area determination unit 303 detects a text area and a photo area in the binary image data generated by the binarization unit 302. As a result, in the example of the input image shown in FIG. 6(a), as shown in FIG. 6(c), the character area information (X, Y, W, H) 521, 522 indicates that it is an inverted character area. Information (hereinafter referred to as inverted character area information) 523 and photo area information (X, Y, W, H) 524 are obtained.

The area determination process by the first area determination unit 303 described above is performed by a known area identification method (for example, Japanese Patent Laid-Open No. 06-068301). Specifically, the outline of the case where area determination is performed on the binary image data shown in FIG. 6(b) is as follows. Black pixel clusters are detected by tracing the contours of black pixels. As a result, black pixel blocks 1 to 6 are obtained as shown in FIG. 6(d). Then, the obtained black pixel clusters are classified into characters, inverted characters, or photographs using at least one of size, shape, and black pixel density. For example, black pixel blocks 1 to 4 whose aspect ratio is close to 1 and whose size is within a predetermined range are determined to be black pixel blocks constituting a character. Further, a pixel block 5 having a rectangular shape and a high black pixel density is determined to be an inverted character. The remaining black pixel blocks 6 are determined to be pixel blocks constituting a photograph. Further, if the distance between black pixel blocks forming a character is within a predetermined distance (for example, 3 pixels), the black pixel blocks are classified into the same group. Then, a circumscribed rectangular area that includes any of the black pixel blocks classified into the same group is determined to be a character area. As a result, black pixel blocks 1 and 2 and black pixel blocks 3 and 4 shown in FIG. 6(d) are determined to be close to each other, and are determined to be character areas. Through the above determination process, a determination result is output that the information 521 and 522 shown in FIG. 6(c) are a text area, the information 523 is an inverted text area, and the information 524 is a photo area.

Next, the second area determination unit 304 performs character extraction processing on the area determined by the first area determination unit 303 to be a character area. Thereby, character area information (x, y, w, h) can be obtained for each character. FIG. 6(e) shows the result of character segmentation processing performed on the character area information 521 to 523 shown in FIG. 6(c). That is, the character area information 521 is separated into unit character area information 541 and 542, the character area information 522 is separated into unit character area information 543 and 544, and the inverted character area information 523 is separated into unit character area information 545 and 546. detected. The above character cutting process separates and detects individual characters by cutting out the circumscribed rectangle of each character as a character cutting rectangle in a character area based on horizontal and vertical projections.

Referring again to FIG. 5, the MMR compression unit 305 inputs the binary image data binarized by the binarization unit 302, and applies data to the area determined to be a character area by the first area determination unit 303. Perform MMR compression. The MMR data that has been subjected to MMR compression is input to the PDF generation unit 310.

The reduction unit 307 reduces the input data (RGB multi-value image data) read by the scanner unit 201 to generate a reduced multi-value image. The generated reduced multilevel image is temporarily stored in the RAM 206. In this embodiment, reduction means resolution conversion to a lower resolution, such as resolution conversion using a bicubic method. The character area filling unit 308 refers to the binary image data binarized by the binarizing unit 302 and the character area information (X, Y, W, H) obtained by the first area determining unit 303. , calculate the average value of the background color within the character area. Further, with reference to the unit character area information (x, y, w, h) obtained from the second area determination unit 304, the calculated average value of the background color is assigned to the unit character area of the reduced multivalued image. That is, a unit character area in the character area of the reduced multi-valued image is filled with the calculated background color to generate a filled-in reduced multi-valued image. This improves the compression rate of the JPEG compression unit 309, which will be described later. A JPEG compression unit 309 performs JPEG compression on the filled-in reduced multivalued image generated by the character area filling unit 308 . The JPEG compressed JPEG data is input to a PDF generation unit 310, which will be described later.

The character color extraction unit 306 extracts character area information (X, Y, W, H) obtained from the first area determination unit 303 and the second area determination unit 304, and unit character area information (x, y, w, See h). Then, while referring to these, the black part of the binary image data generated by the binarization unit 302 and the reduced multilevel image generated by the reduction unit 307 are positioned in correspondence with each other, and the representative color of each character in the character area is Extract.

FIG. 7 is a schematic diagram for explaining the processing of the character color extraction unit 306 of this embodiment. In FIG. 7, reference numerals 801, 802, and 803 indicate 150 dpi reduced multilevel images generated by the reduction unit 307, and reference numerals 804, 805, and 806 indicate 300 dpi reduced multilevel images generated by the binarization unit 302. A binary image is shown. Further, numerals 807, 808, and 809 indicate the character colors of each character in the binary image data (300 dpi), and numerals 810, 811, and 812 indicate the font colors after character color extraction.

Here, the character color 810 refers to the unit character area information (x, y, w, h) of the binary image data 804, and the color value corresponding to the black part of the binary image is set to the reduced multi-value image 801. Get from. FIGS. 8(a) and 8(b) are schematic diagrams for explaining details of this process. FIG. 8A is an enlarged view of a 300 dpi binary image 804. FIG. 8(b) is an enlarged view of the 150 dpi reduced multilevel image 801. Note that since FIGS. 8(a) and 8(b) are illustrated with the same pixel size, the 150 dpi image shown in FIG. 8(b) is the same as that shown in FIG. 8(a). The width and height are 1/2 that of a 300 dpi image. The character color extraction unit 306 obtains color values 1507 corresponding to the positions of black pixels 1503 to 1506 of the binary image shown in FIG. 8(a) in the reduced image shown in FIG. 8(b). In this way, the color values corresponding to all the black pixels of the binary image are obtained, and the average value is calculated.

In FIG. 7, reference numeral 807 indicates the color of the average value of each character in the binary image obtained in this way. As shown in average color 807, the letter "E" has an average color (R, G, B) = (80, 80, 80), and the letter "F" has an average color (R, G, B) = (80, 80, 80). (75, 75, 75). In this way, the average colors of the letters "E" and "F" are generally different. This is because the reduced multilevel image 801 is a reduced version of data read by a scanner. In other words, the color values of both the letter "E" and the letter "F" vary from pixel to pixel, and the halftone state that occurs at the edge of the letter due to reduction differs between the letters "E" and "F". This is because they are different. On the other hand, in order to align the color values for each character, similar colors, for example, within a predetermined luminance difference and color difference range, are replaced with one representative color. In this embodiment, when selecting the one representative color, a method is used in which the average color of characters with a large number of black pixels in the binary image data before average color calculation is selected as the representative color. That is, when comparing the number of pixels that make up the letter "E" and the number of pixels that make up the letter "F," the number of pixels that make up the letter "E" is larger. Therefore, the average color (R, G, B)=(80, 80, 80) of the letter "E" is selected as the representative color. Reference numeral 810 in FIG. 7 indicates that the average color of the letter "E" which is the representative color selected in this way is the representative color (R, G, B) of the letter "F" = (80, 80, 80). Which indicates that.

Similarly, with reference to the unit character area information (x, y, w, h) of the 300 dpi binary image data 805 shown in FIG. 7, the color value corresponding to the black part of the binary image data is reduced by 150 dpi. It is acquired from the multivalued image 802. Reference numeral 808 in FIG. 7 indicates the average color for each character obtained in this manner. As shown in average color 808, the letter "E" has an average color (R, G, B) = (100, 80, 100), and the letter "F" has an average color (R, G, B) = (100, 80, 100). (95, 75, 95). Further, reference numeral 811 in FIG. 7 indicates that the representative colors of the letters "E" and "F" are (R, G, B)=(100, 80, 100). Similarly, with reference to the unit character area information (x, y, w, h) of the 300 dpi binary image data 806 shown in FIG. 7, the color value corresponding to the black part of the binary image data is reduced by 150 dpi. It is acquired from the multivalued image 803. Reference numeral 809 in FIG. 7 indicates the average color for each character obtained in this way. As shown in average color 809, the letter "E" has an average color (R, G, B) = (140, 140, 140), and the letter "F" has an average color (R, G, B) = (140, 140, 140). (135, 135, 135). Further, reference numeral 812 indicates that the representative colors of both the letter "E" and the letter "F" are (R, G, B)=(140, 140, 140). As described above, the text color extraction unit 306 extracts the text color for each text area.

Referring to FIG. 5 again, the text color correction unit 311 corrects the text color obtained by the text color extraction unit 306. Thereby, the color value of the edge portion of the character of the input data (multivalued image data) is influenced by the background color due to reduction in the reduction unit 307, and it is possible to reduce changes in the finally obtained character color. That is, the text color is corrected based on the relationship between the text color and the background color of the text.

One embodiment of the present invention corrects the text color according to the following equation (2).
Text color after correction (R, G, B) = Text color before correction (R, G, B) + {Text color before correction (R, G, B) - Background color (R, G, B)} ×Correction coefficient...(2)

Furthermore, the correction coefficient in equation (2) is calculated using equation (3) below.
Correction coefficient = Number of pixels forming an edge/Number of pixels forming a character...(3)

Here, pixels forming an edge can be detected as follows. FIG. 9 is a diagram schematically showing a window for edge detection. FIG. 10 is a schematic diagram for specifically explaining the correction coefficient. FIG. 3 is a schematic diagram for specifically explaining a correction coefficient. A 3×3 window as shown in FIG. 9 is applied to 300 dpi binary image data (numerals 901, 902, and 903 in FIG. 10). In the 3×3 window, “1” to “9” indicate the positions of pixels of binary image data when the window is applied to the binary image data. In this window, the window position "5" is made to correspond to the pixel of interest for which it is determined whether or not it is an edge in the binary image. Furthermore, the positions "1", "2", "3", "4", "6", "7", "8", and "9" are made to correspond to the pixels around the pixel of interest, respectively. By applying such a window, a pixel with black = "1" in a binary image is set as a pixel of interest, and the pixel of interest is sequentially changed to determine whether or not the pixel of interest constitutes an edge. go. Then, the pixel of interest at position "5" is "black", and the pixel at position "1", "2", "3", "4", "6", "7", "8", "9" If at least one of the surrounding pixels is "white="0", the pixel of interest at position "5" is determined to be a pixel forming an edge.

Note that the above example is an example in which pixels affected by the color of the background are detected as edges when a 300 dpi image is reduced to a 150 dpi image, as shown in FIG. When the reduction rate changes, the range of pixels affected by this background color changes. For example, when reducing a 600 dpi image to a 150 dpi image, 4×4 pixels are converted to 1 pixel. In this case, the pixels affected by the reduction are 3×4 pixels out of 4×4 pixels. Therefore, the window for detecting these pixels as edges has a size of 7×7. In this way, the size of the window used is changed depending on the reduction ratio.

Further, in the above example, edges are detected in a 300 dpi binary image, but they may also be detected in a 300 dpi multilevel image. In this case, the above "black" is represented by (0, 0, 0), and "white" is represented by (255, 255, 255), for example. In the case of inverted characters (reference numeral 903 in FIG. 10), after inverting the black and white of the binary image data based on the inverted character area information obtained from the first area determination unit 303, the window in FIG. By applying this, pixels forming an edge are detected. As described above, pixels forming edges for each character are detected and their number is calculated.

In this embodiment, the correction coefficient is calculated using equation (3) for the following reason. Since the color value of the edge portion of the character is easily influenced by the background color due to reduction in the reduction unit 307, the degree of influence is quantified. In other words, the smaller the character (small point character or character including thin lines), the larger the proportion of pixels forming edges in the character, and the more the character as a whole is more susceptible to the influence of the background color. Therefore, the smaller the character, the larger the correction coefficient. Conversely, the larger the character, the smaller the proportion of the pixels constituting the edge of the character, and the less affected by the background color. Therefore, the larger the character, the smaller the correction coefficient. Note that the correction coefficient may be determined based on the size of the character itself, for example. Specifically, since the width and height of each character are known in the unit character area information obtained by the second area determination unit 304, the correction coefficient can be directly determined using this information.

Regarding the correction coefficient, when the color value of the edge portion is influenced by the background color due to the reduction of the image, the degree of influence is not limited to the ratio of the number of pixels described above. For example, depending on the appearance of the color that changes due to the influence of the background color, it may be a power of the ratio of the number of pixels or a difference in the number of pixels. Furthermore, in addition to the ratio of the number of pixels, the correction coefficients may be weighted according to color differences, such as the color difference between the background color and the text color.

In addition, in the above example, the color change due to reduction was explained using color values (R, G, B) in RGB space, but it is not limited to this. For example, it can also be performed using color values in YCbCr or Lab space. good.

FIG. 10 is a schematic diagram for specifically explaining an example of a correction coefficient. In FIG. 10, reference numerals 901, 902, and 903 indicate 300 dpi binary image data generated by the binarization unit 302. Reference numerals 904, 905, and 906 indicate edge images obtained from 300 dpi binary image data. Reference numerals 907, 908, and 909 indicate correction coefficients.

First, in the case of a small character (a small point character or a character containing thin lines), as shown in binary image data 901, the number of pixels that make up the letter "E" is 110, and the number of pixels that make up the letter "F" is 110. The number of pixels is 90. Therefore, in this case, the number of pixels constituting a character (average for one character) is 100. Next, as shown in edge image 904, the number of pixels forming the edge of the letter "E" is 36, and the number of pixels forming the edge of the letter "F" is 44. Therefore, the average number of pixels forming an edge is 40 per character. From the above, the correction coefficient is 0.4 as calculated by the arithmetic expression for the correction coefficient 907.

Similarly, in the case of large letters, the number of pixels forming the letter "E" is 450, and the number of pixels forming the letter "F" is 350, as shown in binary image data 902. Therefore, the number of pixels constituting a character (average for one character) is 400. Next, as shown in edge image 905, the number of pixels forming the edge of the letter "E" is 85, and the number of pixels forming the edge of the letter "F" is 75. Therefore, the average number of pixels forming an edge is 80 per character. As a result, the correction coefficient becomes 0.2 as calculated by the arithmetic expression in the correction coefficient 908.

In the case of inverted characters, as shown in binary image data 903, after black and white inversion is performed based on the inverted character area information obtained from the first area determination unit 303, the number of pixels composing the character and edges are determined. Calculate the number of constituent pixels. The number of pixels forming the letter "E" is 110, and the number of pixels forming the letter "F" is 90. Therefore, the average number of pixels constituting a character is 100 for each character. Next, as shown in edge image 906, the number of pixels forming the edge of the letter "E" is 36, and the number of pixels forming the edge of the letter "F" is 44. Therefore, the average number of pixels forming an edge is 40 per character. As a result, the correction coefficient becomes 0.4 as calculated by the calculation formula for the correction coefficient 909.

FIG. 11 is a diagram showing the results of correcting the representative colors of extracted characters using equation (2). Assuming that the correction coefficient is 0.4 as described above, the extracted representative color 810 (FIG. 7) ((R, G, B) = (80, 80, 80) as shown in the corrected character color 1001 ) is corrected to representative colors (R, G, B) = (10, 10, 10). In this way, the extracted representative color 810 is corrected to a color closer to the character color ((R, G, B) = (0, 0, 0)) in the input image. As a result, the influence of the background color on the character color due to image reduction can be reduced.

Similarly, as shown in the corrected character color 1002, the representative color 811 (FIG. 7) ((R, G, B) = (100, 80, 100)) is the representative color (R, G, B) = Corrected to (38, 32, 38). Furthermore, as shown in the corrected character color 1003, the representative color 812 (FIG. 7) ((R, G, B) = (140, 140, 140)) is the representative color (R, G, B) = (196, 196, 196).

Referring again to FIG. 5, the PDF generation unit 310 synthesizes the MMR data compressed by the MMR compression unit 305, the text color obtained from the text color correction unit 311, and the JPEG data compressed by the JPEG compression unit 309. Then, by converting the combined data into a PDF format (PDF data), a PDF (highly compressed) is generated.

According to the embodiment described above, the text color in the PDF data in which the corrected text color is combined has a color value closer to the text color in the input image than the text color in the PDF data in which the text color is not combined. .

FIGS. 12(a), 12(b), and 12(c) are flowcharts showing the processing executed by the data processing unit 215 (FIG. 4) described above. Specifically, these flowcharts show each process executed by the data processing unit 215: generation of MMR data, generation of JPEG data, and generation of text color data. Note that a program that executes the processing shown in the flowchart is stored in the ROM 210 or the storage unit 211 in FIG. 4, and is executed by the CPU 205. The CPU 205 can exchange data with the data processing unit 215 using the ImageBus I/F 212, the system bus 216, and the image bus 217. Further, the symbol "S" in the explanation of the flowchart represents a step. This point also applies to the description of the flowcharts below.

<Generation of MMR data>
FIG. 12(a) shows the MMR data generation process. In S401, gray conversion is performed. The CPU 205 executes processing for generating gray multi-value image data from the input data (RGB multi-value image data) read by the scanner unit 201 . The details of this processing are as described above in the explanation of the operation of the gray conversion unit 301. Next, in S402, binarization is performed. The CPU 205 executes processing to generate binary image data from the gray multi-value image data obtained in S401. The details of this processing are as described above in the explanation of the operation of the binarization section 302. Furthermore, in S403, a first region determination is performed. The CPU 205 executes a process of detecting a text area and a photo area from the binary image data generated in S402. The character area information (X, Y, W, H) and inverted character area information obtained here are temporarily stored in the RAM 206. The details of this process are as described above in the explanation of the operation of the first area determining section 303. Next, in S404, a second region determination is performed. The CPU 205 performs character extraction processing on the area determined to be a character area in S403. The unit character area information (x, y, w, h) obtained here is temporarily stored in the RAM 206. The details of this process are as described above in the explanation of the operation of the second region determining section 304. Finally, in S405, MMR compression is performed. The CPU 205 inputs the binary image data binarized in S402, and performs MMR compression on the area determined to be a text area (the text area on the binary image data) in S403. The details of this processing are as described above in the explanation of the operation of the MMR compression section 305.

<Generation of JPEG data>
FIG. 12(c) is a flowchart showing the JPEG data generation process. In S420, the CPU 205 reduces the input data (RGB multi-value image data) to generate a reduced multi-value image. The generated reduced multilevel image is temporarily stored in the RAM 206. The details of this processing are as described above in the explanation of the operation of the reduction unit 307. Next, in S421, blank filling processing for the character area is performed. The CPU 205 refers to the binary image data generated in step S402 and the character area information (X, Y, W, H) obtained in S403, and calculates the average value of the background color in the character area. . Next, referring to the unit character area information (x, y, w, h) obtained in S404, the calculated average value of the background color is assigned to the unit character area of the reduced multivalued image. That is, the unit character area of the reduced multi-valued image is filled with the calculated background color to generate a filled-in reduced multi-valued image. The details of this character area filling process are as described above in the explanation of the operation of the character area filling section 308. Further, in S422, the CPU 205 performs JPEG compression on the hole-filling reduced multivalued image generated in S421. The details of JPEG compression are as described above in the explanation of the operation of the JPEG compression unit 309. The JPEG data generated as described above is temporarily stored in the RAM 206.

<Generation of text color data>
FIG. 12(b) is a flowchart showing the text color data generation process. In S410, the CPU 205 searches for the first character area by referring to the binary image data and the character area information (X, Y, W, H) obtained in S403. Then, in S411, the CPU 205 determines whether the attention area is a character area. If a determination result is obtained that the region of interest is a character region (YES in S411), the process moves to S412. If a determination result is obtained that the attention area is not a character area (NO in S411), the process moves to S414. Next, in S412, the CPU 205 refers to the character area information (X, Y, W, H) obtained in S403 and the unit character area information (x, y, w, h) obtained in S404. . Then, referring to this information, the character color of each character area is extracted while positionally corresponding the black portion of the binary image data and the reduced multivalued image generated in S420. The method for extracting this representative color is as described above in the explanation of the operation of the character color extraction section 306.

Next, in S413, the CPU 205 corrects the character color for each character area obtained in S412. FIG. 13 is a flowchart showing text color correction processing. Note that in this embodiment, the process of the flowchart shown in FIG. 13 is executed for each character area. Furthermore, in a second embodiment to be described later, the process of the flowchart shown in FIG. 13 is executed for each unit character area.

In FIG. 13, first, in S701, the CPU 205 refers to the binary image data and the character area information (X, Y, W, H) obtained in S403, and calculates the background color in the character area. Next, in S702, the CPU 205 performs edge detection from the binary image data generated in S402. Edge detection is as described above in the explanation of the operation of the text color correction section 311. Furthermore, in S703, the CPU 205 calculates a correction coefficient using the binary image data generated in S402 and the edge detection image generated in S702. The calculation of the correction coefficient is as described above in the explanation of the operation of the text color correction section 311. Finally, in S704, the CPU 205 corrects the text color using the text color extracted in S406, the background color calculated in S701, and the correction coefficient calculated in S703. The text color correction is as described above in the explanation of the operation of the text color correction unit 311.

Referring again to FIG. 12, after the text color correction in S413, in S414, the CPU 205 determines whether the search for all character areas has been completed. If it is determined that the search for all character areas has been completed (S414: YES), the process ends. If it is determined that the search for all character areas has not been completed (NO in S414), the process moves to S415. In S415, the CPU 205 searches for the next character area. After the search in S415, the process moves to S411, and the processes from S411 onwards are performed on the next character string area searched in S415. The character color data for each character area generated as described above is temporarily stored in the RAM 206.

<Generation of PDF (highly compressed)>
As described above in the flowcharts shown in FIGS. 12(a), 12(b), and 12(c), MMR data, JPEG data, and text color data are generated, and the CPU 205 converts these data into PDF format. This will generate a PDF (highly compressed).

As described above, the present embodiment corrects the character color after character color extraction in PDF (high compression) based on the color around the character (background color) and the edge information of the character. This makes it possible to reduce changes in the text color, such as brightness and saturation, due to mixing of surrounding colors with the text color.

(Second embodiment)
In the first embodiment described above, a case has been described in which the background color within a character area is calculated and the character colors of character strings within the character area are collectively corrected. In this embodiment, a case will be described in which the character color is corrected for each character. As a result, when the background color changes for each character within the character area or when the size of the characters changes, it becomes possible to perform character color correction suitable for each character. Note that, below, descriptions of processes similar to those in the first embodiment will be omitted.

FIG. 14 is a schematic diagram showing an example of an image in which the background color changes and the character size changes within the same character area. In FIG. 14, reference numeral 1101 indicates input data (multivalued image data). Here, the letter "E" is a capital letter, and the background color is white ((R, G, B) = (255, 255, 255)). The letter "F" is a lowercase letter, and the background color is pink ((R, G, B) = (255, 200, 255)). Reference numeral 1102 indicates a reduced multilevel image obtained by reducing the input data 1101 by a reduction unit (reference numeral 307 in FIG. 5). As described above in the first embodiment, it can be seen that the background color is mixed with the character color at the edge portion of the character. Reference numeral 1103 indicates binary image data obtained by binarizing the input data 1101 by a binarization unit (reference numeral 307 in FIG. 5). The unit character area information (x, y, w, h) of each of the character "E" and the character "F" is shown. The code 1104 indicates the character color of each character in the binary image, and the average color of the character "E" is ((R, G, B) = (50, 50, 50)), and the average color of the character "F" is It shows that the average color of is ((R, G, B) = (95, 75, 95)).

In the first embodiment, in order to align the color values for each character, before correcting the character color, similar colors, for example, colors within a predetermined brightness difference and color difference range, are combined into one representative color. We are planning to replace it. In contrast, in this embodiment, the color is replaced with one representative color after the text color is corrected, not before the text color is corrected. As a result, if there are similar colors in the same character area, it is possible to have the same color value even after character color correction, and if the colors are different (not similar colors), each character This enables text color correction suitable for

FIG. 15 is a schematic diagram showing the result of character color correction for each character using the average color for each character shown in the character color 1104 in FIG. As shown in Figure 15, for the letter "E", background color (R, G, B) = (255, 255, 255) and correction coefficient 0.19 (=85/450) are used, and for the letter "F", , background color (R, G, B)=(255, 200, 255) and a correction coefficient of 0.5 (=44/90). As a result, the character "E" is corrected to the character color (R, G, B) = (11, 11, 11), and the character "F" is corrected to the character color (R, G, B) = (15, 12). , 15).

The font color of the letter “E” (R, G, B) = (11, 11, 11) and the font color of the letter “F” (R, G, B) = (15, 12, 15) are within a predetermined luminance difference (for example, 10). In this embodiment, similar colors are used within this range. Therefore, in order to equalize the color values for each character, they are replaced with one representative color. Comparing the number of pixels that make up "E" and the number of pixels that make up "F", since the number of pixels that make up "E" is large, the average color of the letter "E" (R, G, B) = ( 11, 11, 11) are selected as representative colors.

In this way, according to this embodiment as well, the color of the characters finally obtained by correcting the color of the characters is closer to the color of the characters in the input image data. As a result, it is possible to reduce changes in the text color, such as brightness and saturation, due to the background color being mixed with the text color.

FIG. 16 is a flowchart showing the processing executed by the data processing unit 215 (FIG. 4) according to the present embodiment. A program that executes the processing shown in the flowchart is stored in the ROM 210 or the storage unit 211 in FIG. 4, and is executed by the CPU 205. At this time, the CPU 205 can exchange data with the data processing unit 215 using the ImageBus I/F 212, the system bus 216, and the image bus 217. Below, the process of correcting the character color for each character, which is different from the first embodiment, will be explained with reference to the flowchart shown in FIG. 16. In this description, the processes shown in FIG. 12 and FIG. 13 described above in the first embodiment will be referred to as appropriate. Further, the processes of S410, S411, S414, and S415 shown in FIG. 16 are the same as the processes of S410, S411, S414, and S415 of FIG. 12(b), so the description thereof will be omitted.

First, in S1301, the CPU 205 refers to the unit character area information (x, y, w, h) obtained in S404 for the area determined to be a character area in S411, and selects the first unit character area. Execute processing to search for. Next, in S1302, the CPU 205 refers to the unit character area information (x, y, w, h) obtained in S404. Then, referring to this information, a process is executed to extract the character color for each unit character area in the character area while positionally corresponding the black part of the binary image data and the reduced multivalued image generated in S420. . The representative color extraction method is as described above in the operation of the character color extraction unit 306.

Next, in S1303, the CPU 205 corrects the character color obtained in S1302. In the first embodiment, this correction process is performed for each character area. In this embodiment, the process shown in FIG. 16 is executed for each unit character area.

In S1304, the CPU 205 executes processing to determine whether the search for all unit character areas has been completed. If it is determined that the search for all unit character areas has been completed (YES in S1304), the process moves to S1305. If it is determined that the search for all unit character areas has not been completed (NO in S1304), the process moves to S1306. In S1306, the CPU 205 searches for the next unit character area. After the search in S1306, the process moves to S1302, and the processes from S1302 onwards are performed on the next unit character string area searched in S1306.

In S1305, the CPU 205 executes processing to replace the corrected text color obtained in S1303 with a representative color. If a color is similar within the same character area, for example, within a predetermined brightness difference and color difference range, it is replaced with one representative color.

As described above, by correcting the font color for each character, if the background color changes within the character area or the font size changes, the font color can be corrected to suit each character. becomes. In addition, if there are similar colors in the same character area, it is possible to maintain the same color value even after character color correction, and if the colors are different (not similar colors), the color value suitable for each character can be set. Text color correction becomes possible.

(Other embodiments)
The present invention provides a system or device with a program that implements one or more functions of the embodiments described above via a network or a storage medium, and one or more processors in a computer of the system or device reads and executes the program. This can also be achieved by processing. It can also be realized by a circuit (for example, ASIC) that realizes one or more functions.

306 Text color extraction unit 307 Reduction unit 310 PDF generation unit 311 Text color correction unit

Claims (6)

Reduction means for converting the first image into a second image having a lower resolution than the first image;
extraction means for extracting a character color in a character area of the second image obtained by the reduction means;
The character color extracted by the extraction means is determined based on the relationship between the color value of the character color and the color value of the background color in the character area of the second image , and the number of pixels forming the edge of the character in the character area. , a correction means for correcting based on the number of pixels constituting the character ;
generation means for generating a compressed image in which the first image is compressed as an image including the character area whose character color has been corrected by the correction means;
An image processing device comprising: The correction means corrects the text color by multiplying a value obtained from the relationship between the color value of the text color and the color value of a background color in the text area of the second image by a correction coefficient. The image processing apparatus according to claim 1. The correction means is
Text color after correction = Text color before correction + (Text color before correction - Background color) x Correction coefficient Here, the correction coefficient is determined by 3. The image processing apparatus according to claim 2 , wherein the character color is corrected by an equation expressed as: (number of pixels constituting a character)/(number of pixels constituting a character). The generating means generates, as the compressed image,
MMR data generated from the binary image based on information on the character area obtained by area determination on the binary image generated from the first image;
JPEG data generated from the second image;
the corrected text color;
The image processing apparatus according to any one of claims 1 to 3 , wherein the image processing apparatus generates PDF data based on. a reduction step of converting the first image into a second image having a lower resolution than the first image;
an extraction step of extracting a character color in a character area of the second image obtained in the reduction step;
The character color extracted in the extraction step is determined based on the relationship between the color value of the character color and the color value of the background color in the character area of the second image, and the number of pixels forming the edge of the character in the character area. and a correction step of correcting based on the number of pixels forming the character,
a generation step of generating a compressed image in which the first image is compressed as an image including the character area whose character color has been corrected in the correction step;
An image processing method comprising : A program for causing a computer to execute the image processing method according to claim 5 .

Images