JP2008099149A - Image processor, image processing method and image processing program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To easily and highly accurately detect the colors of characters. <P>SOLUTION: An MFP includes: a character region extraction part 310 for extracting the character region of a character attribute where the character is indicated from color images; a thinning part 331 for thinning the character included in the character region; a character color determination part 330 for determining a character color on the basis of the color of the pixels of the color image corresponding to a plurality of character center pixels constituting the thinned character; a noise elimination part 320 for converting the color image to a binary image by binarizing it; a character region encoding part 340 for encoding the character region of the binary image and generating code data; and an integration part 370 for relating the generated code data, the determined character color and the position in the color image of the character region. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an image processing apparatus, an image processing method, and an image processing program, and more particularly, to an image processing apparatus, an image processing method, and an image processing program for efficiently compressing a color image.

  In recent years, with the progress of computerization of information, there is an increasing demand for storing or transmitting documents as electronic data instead of paper. Further, the document is often expressed in color, and it is desired that the electronic data is also a color image. On the other hand, since a color image has a larger data amount than a monochrome image, the electronic data of the color image is compressed and the compressed data is stored or transmitted.

  However, when characters are represented in a color image, there is a problem that if the image is compressed by an irreversible method suitable for compression of a photograph or the like, the characters cannot be read in the expanded color image. To cope with this problem, a color image is separated into a character area where characters are represented and an area other than the character area, the character area of the color image is compressed by a lossless compression method, and an area other than the character area is compressed. A technique for compressing with a lossy compression method is described in Japanese Patent Laid-Open No. 2000-350040 (Patent Document 1) or Japanese Patent Laid-Open No. 5-130426 (Patent Document 2). Furthermore, when characters are represented in a plurality of colors in a color image, the color image can be efficiently compressed by reversibly compressing an image in which character regions are binarized for a plurality of colors.

However, due to the characteristics of the scanner that reads the document, the color image data output by the scanner has a different color between the pixel at the center of the character and the pixel at the outline of the character. In order to detect this with high accuracy, complicated processing must be performed, and there is a problem that processing takes time.
JP 2001-31725 A JP 2001-31726 A

  The present invention has been made to solve the above-described problems, and one of the objects of the present invention is to provide an image processing apparatus, an image compression method, and an image capable of easily and accurately detecting the color of a character. It is to provide a compression program.

  Another object of the present invention is to provide an image processing apparatus with improved accuracy for extracting a region where characters are represented from an image.

  In order to achieve the above-described object, according to one aspect of the present invention, an image processing device includes: a character area extracting unit that extracts a character area having a character attribute representing a character from a color image; and a character included in the character area. Thinning means for thinning the character, character color determining means for determining the character color based on the color of the pixel of the color image corresponding to the plurality of character center pixels constituting the thinned character, and two color images. By binarization, binarization means for converting to a binary image, character area encoding means for encoding a character area of the binary image and generating code data, and generated code data are determined Association means for associating the character color with the position of the character region in the color image.

  According to this aspect, the character included in the character region extracted from the color image is thinned, and the character is based on the color of the pixel of the color image corresponding to the plurality of character center pixels constituting the thinned character. The color is determined. Then, the character area of the binary image obtained by binarizing the color image is encoded, and the code data, the determined character color, and the position of the character area in the color image are associated with each other. Since the character area of the binary image obtained by binarizing the color image is encoded, the entire character can be binarized as a character of the same color. Further, since the character color is determined based on the color of the pixel of the image data corresponding to the plurality of character center pixels constituting the thinned character, the character color can be accurately detected. As a result, it is possible to provide an image processing apparatus that can easily and accurately detect the color of characters in an image.

  Preferably, the thinning means is continuous in the sub-scanning direction intersecting with the main scanning direction and main scanning direction detecting means for extracting a central pixel from black pixels continuously arranged in the main scanning direction in the binary image. Sub-scanning direction detecting means for extracting the central pixel from the black pixels arranged in the same manner, the central pixel extracted by the main scanning direction detecting means, and the central pixel extracted by the sub-scanning direction detecting means Determining means for determining the center pixel.

  According to this aspect, since the center pixel is extracted as the character center pixel from the continuously arranged black pixels in the main scanning direction and the sub-scanning direction, the character center pixel can be easily extracted. , Processing speed can be increased.

  Preferably, the thinning unit includes a main scanning direction detecting unit that extracts at least one pixel excluding a predetermined number of pixels from both ends of a plurality of black pixels continuously arranged in the main scanning direction as a first candidate pixel. Sub-scanning direction detecting means for extracting at least one pixel excluding a predetermined number of pixels from both ends of a plurality of black pixels continuously arranged in the sub-scanning direction intersecting with the main scanning direction; Determining means for determining a pixel that is both a first candidate pixel and a second candidate pixel as a character center pixel.

  According to this aspect, at least one pixel obtained by removing a predetermined number of pixels from both ends of a plurality of black pixels continuously arranged in the main scanning direction, and a plurality of pixels arranged continuously in the sub-scanning direction Since at least one pixel excluding a predetermined number of pixels from both ends of the black pixel is used as the character center pixel, the character center pixel can be easily extracted, and the processing speed can be increased.

  Preferably, the character color determining means includes a classification means for classifying each of the plurality of character center pixels into one of a plurality of hue classes based on the color before thinning of each of the plurality of character center pixels, and the frequency is first. Noise removal means for removing pixels classified into a hue class that is equal to or less than the threshold value as noise.

  Preferably, the character color determination means includes a classification means for classifying each of the plurality of character center pixels into one of a plurality of hue classes based on a color before thinning of each of the plurality of character center pixels, and a plurality of hue classes. Selecting a hue class whose frequency exceeds a first threshold value as a representative hue, and the character center pixel classified into the hue class of the selected representative hue among a plurality of character center pixels The character color is determined based on the color before thinning.

  Preferably, the selection unit includes a hue integration unit that selects, as a representative hue, a hue class having the highest frequency among consecutive hue classes when two or more hue classes having a frequency exceeding the first threshold value are consecutive.

  Preferably, the character color determining means determines a plurality of character colors respectively corresponding to the plurality of representative hues when the selection means selects a plurality of representative hues, and the character area encoding means In this case, by extracting a pixel in which the color of the pixel of the color image corresponding to the black pixel approximates the character color corresponding to the representative hue from among the black pixels included in the character region of the binary image, the representative hue Character color binarized data generating means for generating a character color binarized image corresponding to the character color, and a plurality of character color binarized data respectively corresponding to the determined character colors A character color encoding unit that generates code data, and the association unit includes, for each of the plurality of determined character colors, the character color, the character color code data corresponding to the character color, and a character Position in the image data of the area Including the character color by association means for associating the door.

  Preferably, the character color determination means includes a classification means for classifying each of the plurality of character center pixels into one of a plurality of hue classes based on a color before thinning of each of the plurality of character center pixels, and a plurality of hue classes. And an attribute changing means for changing the attribute of the character area to an attribute other than the character attribute based on the frequency distribution.

  According to another aspect of the present invention, an image processing apparatus includes a character area extracting unit that extracts a character area having a character attribute representing a character from a color image, and a photograph / figure representing a photograph or a figure from the color image. A photo / figure area extracting means for extracting a photograph / figure area of an attribute, a thinning means for thinning a character included in the character area, and thinning each of a plurality of character center pixels constituting the thinned character. Based on the previous color, the character area attribute is changed to an attribute other than the character attribute based on the classification means for classifying each of the plurality of character center pixels into one of a plurality of hue classes and the frequency distribution of the plurality of hue classes Attribute changing means.

  According to this aspect, the attribute of the character area is changed to an attribute other than the character attribute based on the frequency distribution in a plurality of hue classes of the color before thinning of each character center pixel. For this reason, even when the area where the photograph or the figure is represented is erroneously determined as the character area, the attribute of the area can be appropriately determined. As a result, it is possible to provide an image processing apparatus with improved accuracy for extracting a region where characters are represented from an image.

  According to still another aspect of the present invention, an image processing method includes a step of extracting a character region having a character attribute representing a character from a color image, a step of thinning a character included in the character region, and a thinning step Determining a character color based on the color of a pixel of a color image corresponding to a plurality of character center pixels constituting the character and converting the color image into a binary image by binarizing the color image Encoding the character region of the binary image and generating code data, associating the generated code data, the determined character color, and the position of the character region in the color image; including.

  According to this aspect, it is possible to provide an image processing method capable of easily and accurately detecting the color of characters in an image.

According to still another aspect of the present invention, an image processing method includes a step of extracting a character area having a character attribute representing a character from a color image, and a photo / graphic attribute representing a photograph or figure from the color image. A step of extracting a photo / figure region, a step of thinning characters included in the character region, and a plurality of characters based on the color before thinning of each of a plurality of character center pixels constituting the thinned character Classifying each central pixel into one of a plurality of hue classes;
Changing the attribute of the character region to an attribute other than the character attribute based on the frequency distribution of a plurality of hue classes.

  If this aspect is followed, the image processing method which improved the precision which extracts the area | region where the character was represented from the image can be provided.

  According to still another aspect of the present invention, an image processing program extracts a character region having a character attribute representing a character from a color image, thinning characters included in the character region, and thinning Determining a character color based on the color of a pixel of a color image corresponding to a plurality of character center pixels constituting the character and converting the color image into a binary image by binarizing the color image Encoding the character region of the binary image and generating code data, associating the generated code data, the determined character color, and the position of the character region in the color image; Is executed on the computer.

  If this aspect is followed, the image processing program which can detect the color of the character in an image easily and with high precision can be provided.

According to still another aspect of the present invention, an image processing program extracts a character region having a character attribute representing characters from a color image, and a photo / graphic attribute representing a photograph or figure from a color image. A step of extracting a photo / figure region, a step of thinning characters included in the character region, and a plurality of characters based on the color before thinning of each of a plurality of character center pixels constituting the thinned character Classifying each central pixel into one of a plurality of hue classes;
And causing the computer to execute a step of changing the attribute of the character region to an attribute other than the character attribute based on the frequency distribution of the plurality of hue classes.

  If this aspect is followed, the image processing program which improved the precision which extracts the area | region where the character was represented from the image can be provided.

  Embodiments of the present invention will be described below with reference to the drawings. In the following description, the same parts are denoted by the same reference numerals. Their names and functions are also the same. Therefore, detailed description thereof will not be repeated.

  FIG. 1 is a diagram showing an overall outline of an image forming system according to one embodiment of the present invention. Referring to FIG. 1, an image forming system 1 includes an MFP (Multi Function Peripheral) 100 as an image processing apparatus connected to a network 2, and computers 3, 3 A, 3 B, and 3 C, each connected to the MFP 100. Digital camera 4, digital video camera 5, and portable information terminal 6.

  MFP 100 includes a scanner for reading a document, an image forming unit for forming an image on a recording medium such as paper based on image data, and a facsimile, and includes an image reading function, a copying function, and a facsimile transmission / reception function.

  Computers 3, 3 </ b> A, 3 </ b> B, and 3 </ b> C are general personal computers, and a printer driver program for controlling MFP 100 is installed. Computers 3, 3 </ b> A, 3 </ b> B, and 3 </ b> C output print data or image data generated when they execute an application program to MFP 100. MFP 100 performs image processing on print data or image data input from portable information terminal 6, or forms an image on a recording medium based on the print data or image data. Furthermore, the computers 3, 3A, 3B, 3C also function as image processing apparatuses.

  The digital camera 4 and the digital video camera 5 store captured still images or moving images, and output image data of the still images or moving images to the MFP 100 to which they are connected. The MFP 100 performs image processing on the input still image data, and forms a still image on a recording medium such as paper based on the still image data. When image data of a moving image is input, image data of one frame image is processed, and an image is formed on a recording medium based on the image data of one frame image. As with the computers 3, 3 A, 3 B, 3 C, the portable information terminal 6 outputs data or image data generated by executing an application program to the MFP 100. MFP 100 performs image processing on data or image data input from portable information terminal 6 and forms an image on a recording medium based on the image data.

  The network 2 is a local area network (LAN), and the connection form may be wired or wireless. The network 2 is not limited to a LAN, and may be a wide area network (WAN), the Internet, or the like.

  FIG. 2 is a perspective view showing the appearance of the MFP. Referring to FIG. 2, MFP 100 includes an automatic document feeder (ADF) 10, an image reading unit 20, an image forming unit 40, and a paper feeding unit 50. The ADF 10 handles a plurality of documents mounted on the document table, and sequentially conveys them to the image reading unit 20 one by one. The image reading unit 20 optically reads image information such as photographs, characters, pictures, and the like from a document and acquires image data. When the image data is input, the image forming unit 40 forms an image on a sheet based on the image data. The image forming unit 40 forms an image using toners of four colors, cyan, magenta, yellow, and black. The paper feed unit 50 stores paper and supplies the stored paper to the image forming unit 40 one by one. MFP 100 includes an operation panel 9 on the upper surface thereof.

  FIG. 3 is a block diagram illustrating an example of a hardware configuration of the MFP. Referring to FIG. 3, MFP 100 includes main circuit 101, facsimile unit 60, communication control unit 61, ADF 10, image reading unit 20, image processing unit 30, image forming unit 40, and paper feeding unit. 50. The main circuit 101 includes a central processing unit (CPU) 111, a RAM (Random Access Memory) 112 used as a work area of the CPU 111, a ROM (Read Only Memory) 113 for storing programs executed by the CPU 111, and the like. A display unit 114, an operation unit 115, a hard disk drive (HDD) 116 as a mass storage device, and a data communication control unit 117. The CPU 111 is connected to the display unit 114, the operation unit 115, the HDD 116, and the data communication control unit 117, and controls the entire main circuit 101. The CPU 111 is connected to the facsimile unit 60, the communication control unit 61, the ADF 10, the image reading unit 20, the image processing unit 30, the image forming unit 40, and the paper feeding unit 50, and controls the entire MFP 100.

  The display unit 114 is a display device such as a liquid crystal display (LCD) or an organic ELD (Electro Luminescence Display), and displays an instruction menu for the user, information about acquired image data, and the like. The operation unit 115 includes a plurality of keys, and accepts input of various instructions, data such as characters and numbers by user operations corresponding to the keys. The operation unit 115 includes a touch panel provided on the display unit 114. The display unit 114 and the operation unit 115 constitute the operation panel 9.

  The data communication control unit 117 includes a LAN terminal 118 that is an interface for communicating with a communication protocol such as TCP (Transmission Control Protocol) or FTP (File Transfer Protocol), and a serial communication interface terminal 119. The data communication control unit 117 transmits / receives data to / from an external device connected to the LAN terminal 118 or the serial communication interface terminal 119 in accordance with an instruction from the CPU 111.

  When a LAN cable for connecting to the network 2 is connected to the LAN terminal 118, the data communication control unit 117 is connected to another MFP, scanner, printer, or computer 3, 3A, which is connected via the LAN terminal 118. Communicate with 3B, 3C. When a device is connected to the serial communication interface terminal 119, the data communication control unit 117 communicates with a device connected to the serial communication interface terminal 119, for example, the digital camera 4, the digital video camera 5, or the portable information terminal 6. Communicate and input / output image data. The serial communication interface terminal 119 can be connected to a memory card 119A incorporating a flash memory. The CPU 111 controls the data communication control unit 117 to read a program to be executed by the CPU 111 or the image processing unit 30 from the memory card 119A, and stores the read program in the RAM 112 and executes it.

  A recording medium for storing a program to be executed by the CPU 111 or the image processing unit 30 is not limited to the memory card 119A, but a flexible disk, a cassette tape, an optical disk (CD-ROM (Compact Disc-Read Only Memory) / MO). (Magnetic Optical Disc / MD (Mini Disc) / DVD (Digital Versatile Disc)), IC card (including memory card), optical card, mask ROM, EPROM (Erasable Programmable ROM), EEPROM (Electrical ROM, etc.) Further, the CPU 111 may be programmed from a computer connected to the Internet. The program is downloaded and stored in the HDD 116, or the computer connected to the Internet writes the program in the HDD 116, and the program stored in the HDD 116 is loaded into the RAM 112 and executed by the CPU 111 or the image processing unit 30. The program here includes not only a program that can be directly executed by the CPU 111 or the image processing unit 30 but also a source program, a compressed program, an encrypted program, and the like.

  The communication control unit 61 is a modem for connecting the CPU 111 to a PSTN (Public Switched Telephony Networks) 7. The MFP 100 is assigned a telephone number in the PSTN 7 in advance. When a call is made from the facsimile apparatus connected to the PSTN 7 to the telephone number assigned to the MFP 100, the communication control unit 61 detects the call. When the communication control unit 61 detects a call, the communication control unit 61 establishes a call and causes the facsimile unit 60 to communicate.

  The facsimile unit 60 is connected to the PSTN 7 and transmits facsimile data to the PSTN 7 or receives facsimile data from the PSTN 7. The facsimile unit 60 outputs the received facsimile data to the CPU 111. The facsimile unit 60 converts the received facsimile data into print data that can be printed by the image forming unit 40 and outputs the print data to the image forming unit 40. As a result, the image forming unit 40 prints the facsimile data received by the facsimile unit 60 on a recording sheet. The facsimile unit 60 converts the image data stored in the HDD 116 into facsimile data, and outputs the facsimile data to a facsimile machine connected to the PSTN 7 or another MFP. As a result, the data stored in HDD 116 can be output to a facsimile machine or another MFP. The facsimile data is included in the image data. As described above, the MFP 100 has a facsimile transmission / reception function.

  The MFP 100 has a scanner function, and image data that the image reading unit 20 reads and outputs a document is temporarily stored in the RAM 112. The image forming unit 40 forms an image on a recording sheet such as paper based on image data output by the image reading unit 20 reading a document. Therefore, the MFP 100 has a copy function. Further, the data communication control unit 117 receives image data received from any of the other PCs 3, 3A, 3B, 3C connected to the network 2, or the digital camera 4 connected to the serial communication interface terminal 119, the digital video. Image data received from either the camera 5 or the portable information terminal 6 is temporarily stored in the RAM 112. Then, an image is formed on a recording sheet such as paper based on the image data temporarily stored in the RAM 112. Therefore, MFP 100 has a print function.

  The image processing unit 30 performs image processing on the image data stored in the RAM 112. When the MFP 100 activates the scanner function or the copy function, the image processing unit 30 performs image processing on image data output by the image reading unit 20 reading a document. Further, when MFP 100 activates the facsimile function, image processing unit 30 performs image processing on facsimile data (image data) received by facsimile unit 60. Further, the image processing unit 30 is connected to the image data received by the data communication control unit 117 from any of the other PCs 3, 3 A, 3 B, 3 C, scanner, and MFP connected to the network 2, or to the serial communication interface terminal 119. Image data received from any of the connected digital camera 4, digital video camera 5, and portable information terminal 6 is subjected to image processing.

  FIG. 4 is a functional block diagram illustrating an example of functions of the image processing unit. Referring to FIG. 4, an image processing unit 30 performs preprocessing on image data and outputs a brightness image, and a character region extraction unit 310 that extracts a character region in which characters are represented from the brightness image. A noise removing unit 320 that removes noise from the character region image, a character color determining unit 330 that determines the color of the character included in the character region, and a character region encoding unit 340 that encodes the character region image. A photo / figure region extraction unit 350 that extracts a photo attribute and a graphic attribute region from the brightness image, a photo / figure region encoding unit 360 that encodes a photo attribute and a graphic attribute region, and encoded data And an integration unit 370 for integration.

  In the present embodiment, the image data input to the image processing unit 30 is a 24-bit value in which one pixel includes three values (8 bits) of red (R), green (G), and blue (B). Is a color image.

  The pre-processing unit 301 performs background removal processing for removing the background portion of the color image, lightness conversion processing for converting the color image from which the background is removed into a brightness image, and smoothing processing for smoothing the brightness image. The processed brightness image is output to the character region extraction unit 310 and the photograph / figure region extraction unit 350. The background removal process removes a background color having a low density in a color image and extends the contrast. Thereby, the tendency of a color image can be adjusted. In the brightness conversion process, brightness (8 bits) is calculated based on each pixel value (24 bits) of the color image, and a brightness image is generated. The brightness image is a grayscale image in which one pixel has an 8-bit value. The brightness image is generated because the brightness / image area extraction unit 350 and the character area extraction unit 310 set the brightness image as a processing target. In the smoothing process, the brightness image is smoothed. Since the noise is removed by the smoothing, it is possible to improve the accuracy of extracting a region in the photo / figure region extracting unit 350 and the character region extracting unit 310.

  The main purpose of the photograph / drawing area extraction unit 350 is to extract the area where the figure is represented and the area where the photograph is represented from the color image including the area where the character, the figure, the photograph and the table are represented. The reason why the region where the figure is represented and the region where the photograph is represented is extracted because these regions rarely include characters. If character extraction processing is performed on an area where figures and photographs that contain few characters are included, a part of the area where figures and photographs are displayed will be erroneously determined as characters. This is to avoid erroneous determination. For this reason, in the present embodiment, the photo / figure region extraction unit 350 does not input the brightness data of the region determined to be a figure attribute or a photo attribute to the character region extraction unit 310. For the ruled lines in the table, the ruled line is removed by the character area extracting unit 310, the ruled line position information is transmitted to the photo / figure region encoding unit 360, and the ruled lines are encoded by the photo / figure region encoding unit 360.

  The photograph / drawing region extraction unit 350 binarizes the brightness image and generates a binary image. Then, the generated binary image is labeled. Labeling is a process of extracting a rectangle circumscribing a set of a plurality of adjacent pixels among black pixels having a value of “1” in a binary image as a block. The reason why the block is rectangular is that, in general, a photograph, a figure, and the like are often displayed on a document as a rectangular image. The block extracted by labeling is associated with position information indicating its position in the image data.

  Further, the photograph / figure region extraction unit 350 calculates the size of the extracted block and the feature amount of each brightness image, and determines the attribute of the block. When a plurality of blocks are extracted, the attribute of each of the plurality of blocks is determined. The attribute specifies the object represented in the block of the brightness image, and includes a character attribute represented by a character, a figure attribute represented by a graphic, a photo attribute represented by a photograph, and a table attribute. Then, the photograph / figure region extraction unit 350 outputs the position information of the block of the photograph attribute or the diagram attribute to the photograph image encoding unit 360 and the character region extraction unit 310. The position information is, for example, the coordinates of the diagonal pixels of the block in the image data.

  FIG. 5 is a functional block diagram showing detailed functions of the character area extraction unit. Referring to FIG. 5, the character region extraction unit 310 extracts an edge from the brightness image, corrects the inclination of the edge image from the edge image generation unit 311 that generates the edge image, and a line composed of the extracted edges. An inclination correction unit 312, a ruled line removal unit 313 that removes a ruled line from the edge image whose inclination is corrected, a candidate block extraction unit 314 that extracts a block that is a candidate for a character region from the edge image, and an edge image of the block A determination unit 315 that determines whether or not a character attribute is present, a character region determination unit 316 that determines a block that is a character attribute as a character region, and a first that changes a block that is determined not to be a character attribute to a region of a photo attribute Attribute change unit 317.

  The edge image generation unit 311 receives the brightness image output from the preprocessing unit 301 and generates an edge image by filtering the brightness image using, for example, a differential filter. The edge image includes, as edge pixels (value is “1”), in the brightness image, the pixel value has a predetermined difference from the values of surrounding pixels. The inclination correction unit 312 rotates the edge image so that the line formed by the extracted edge pixels is horizontal or vertical. Thereby, the detection accuracy of the ruled line by the next ruled line removing unit 313 is improved. The ruled line removal unit 313 detects a ruled line from the edge image and removes the ruled line from the edge image. This is to improve the accuracy of character determination in the determination unit 315. Further, the brightness data around the part corresponding to the ruled line of the edge image is examined, the ruled line area is determined, and the position information of the ruled line is output to the photo / figure area encoding unit 360.

  The candidate block extraction unit 314 extracts candidate blocks that are candidates for character regions from the edge image. Specifically, the line spacing is first excluded from the processing target. Between the lines, the pixel values in the main scanning direction (X-axis direction) of the edge image are all “0” lines. By connecting neighboring edge pixels (pixels having a value of “1”) with pixels other than between the rows of the edge image, a rectangle circumscribing the set of connected edge pixels is extracted as a candidate block. The candidate block corresponds to a circumscribed rectangle of a character string in one line in the image.

  The determination unit 315 determines whether or not the extracted candidate block has a character attribute. The determination method may be determined based on, for example, the aspect ratio of the candidate block, the length in the horizontal direction, the length in the vertical direction, the ratio of edge pixels in the edge image of the candidate block, and the like. When a plurality of candidate blocks are extracted by the candidate block extraction unit 314, the determination unit 315 determines each of the plurality of candidate blocks.

  The character region determination unit 316 determines the candidate block determined as the character attribute by the determination unit 315 as the character region, and outputs position information indicating the position of the character region in the image data to the noise removal unit 320. The first attribute changing unit 317 changes a candidate block that has not been determined to be a character attribute by the determining unit 315 to a photo area, and sends position information indicating the position of the candidate block in the image data to the photo / figure area encoding unit 360. Output.

  FIG. 6 is a functional block diagram showing detailed functions of the noise removing unit 320. Referring to FIG. 6, noise removal unit 320 receives character region position information from character region extraction unit 310 and brightness image from preprocessing unit 301. The noise removing unit 320 includes a lightness histogram generating unit 321 for generating a lightness histogram based on the pixel values of a plurality of pixels included in the character area of the lightness image, and a halftone dot / background removal for removing noise and background. Part 323 and a second attribute changing part 325 for changing the attribute of the character area to a photo attribute.

  Halftone noise has low brightness. For this reason, the halftone dot / background removal unit 323 removes pixels having lightness lower than a predetermined threshold among a plurality of pixels included in the character area of the lightness image as halftone noise. Specifically, the pixel value of a pixel whose brightness is lower than the threshold value is set to “0”. Furthermore, when a character is represented in the character area of the brightness image, the pixel included in the character area is either a pixel constituting the character or a pixel constituting the background. The brightness differs between the character and the background. For this reason, in the brightness histogram, the frequency increases with the brightness of the character, and the frequency increases with the brightness of the background. In the case of including characters of a plurality of colors, the frequency increases with the brightness of the pixels constituting each character of the plurality of colors. In addition, the background has high lightness and the characters have low lightness. Therefore, the noise removal unit 320 sets a threshold for background removal between the lightness of the character and the lightness of the background from the frequency of the histogram, and binarizes the lightness image to generate a binary image. Specifically, among the plurality of pixels included in the character area of the brightness image, the pixel value of the pixel having a brightness higher than the threshold for background removal is set to “0”, and the values of the other images are set to “1”. " The halftone dot / background removal unit 323 outputs the character region position information and the binary image to the character color determination unit 330 and the character region encoding unit 340.

  The halftone dot / background removal unit 323 may not be able to set a threshold for background removal. This is a case where a photograph or a figure is represented in the character area of the brightness image. Therefore, the halftone / background removal unit 323 outputs an error signal to the second attribute change unit when the threshold for background removal cannot be set. When an error signal is input from the halftone dot / background removal unit 323, the second attribute change unit 325 changes the attribute of the character area to a photo area, and displays position information indicating the position of the character area in the image data. It outputs to the figure area | region encoding part 360. FIG.

  FIG. 7 is a functional block diagram showing detailed functions of the character color determining unit. Referring to FIG. 7, character color determination unit 330 receives a binary image and position information from noise removal unit 320, and receives a color image. The character color determination unit 330 includes a thinning unit 331 that narrows the line width of characters included in the binary image, a hue histogram generation unit 332 that generates a hue histogram from pixel values of the thin lines in the color image, and noise hue removal. A section 333, an adjacent hue integration section 334, a third attribute change section 335, a representative hue selection section 336, a fourth attribute change section 337, and a character color calculation section 338.

  The thinning unit 331 extracts the central pixel from the black pixels continuously arranged in the main scanning direction in the binary image as the central pixel in the main scanning direction, and continuously in the sub-scanning direction intersecting with the main scanning direction. A central pixel is extracted from the arranged black pixels as a central pixel in the sub-scanning direction. Then, the extracted central pixel in the main scanning direction and the central pixel in the sub-scanning direction are determined as character central pixels constituting the character. FIG. 8 is a diagram illustrating an example of a binary image including the character “A”. Pixels constituting the character are indicated by hatching. FIG. 9A is a diagram illustrating an example of the central pixel in the main scanning direction. The central pixel in the main scanning direction extracted from the binary image shown in FIG. 8 is indicated by hatching. FIG. 9B is a diagram illustrating an example of the central pixel in the sub main scanning direction. The center pixel in the sub-scanning direction extracted from the binary image shown in FIG. 8 is indicated by hatching. FIG. 9C is a diagram illustrating an example of the character center pixel. The character center pixel includes the main scanning direction center pixel shown in FIG. 9A and the sub-scanning direction center pixel shown in FIG. 9B, and is hatched.

  Returning to FIG. 7, the thinning unit 331 outputs a binary image in which the value of the character center pixel is “1” and the values of the other pixels are “0” to the hue histogram generation unit 332. As described above, because the color of the outline of the non-black character in the image that it outputs is different from the color of the center of the character, the character color is determined from the color that forms the center of the character. It is to do. Since the character center pixel can be extracted by a simple process of extracting the center pixel in the main scanning direction of the character and the center pixel of the sub-scanning method, the processing speed is improved. Furthermore, since the number of pixels to be processed is reduced, the speed of subsequent processing can be increased.

  The central pixel may be a pixel excluding the outline portion of the character. In this case, the thinning unit 331 extracts at least one pixel obtained by removing a predetermined number of pixels from both ends of a plurality of black pixels continuously arranged in the main scanning direction in the binary image as a first candidate pixel, At least one pixel obtained by removing a predetermined number of pixels from both ends of a plurality of black pixels continuously arranged in the sub-scanning direction is extracted as a second candidate pixel. Then, the pixel that is both the first candidate pixel and the second candidate pixel is determined as the character center pixel. The predetermined number may be determined by the thickness of the character and the reading resolution of the scanner. Even in this case, since the character center pixel can be extracted by a simple process of removing the pixels in the outline portion of the character, the processing speed is improved. Furthermore, since the number of pixels to be processed is reduced, the speed of subsequent processing can be increased.

  The hue histogram generation unit 332 generates a hue histogram based on the pixel value of the color image of the pixel having the value “1” in the binary image input from the thinning unit 331. In other words, the thinned character (character center pixel) is classified into one of a plurality of hue classes based on the color before thinning of the thinned character (character center pixel). Specifically, the pixel value of the pixel in the color image corresponding to the pixel to be processed is acquired from the position of the pixel to be processed in the binary image and the position information of the binarized image. Since the pixel value of the pixel in the color image has an RGB value, the RGB value is converted into the HSL color system. Then, a hue histogram is generated based on the hue (H). However, when the saturation (S) is lower than a predetermined value, the lightness (L) is compared with a threshold value and classified into either a white or black class. In this case, the hue histogram includes a class of 32 sections obtained by dividing the hue into 32 equal parts, a white class, and a black class. FIG. 10 shows an example of the class of the hue histogram.

  The noise hue removing unit 333 sets the first threshold value based on the maximum frequency in the hue histogram generated by the hue histogram generating unit 332, and the pixels distributed to the frequency class equal to or lower than the first threshold value. It is regarded as noise and the pixel is excluded from the processing target. By removing noise, it is possible to improve the accuracy of character color detection. FIG. 11 is a first diagram illustrating an example of a hue histogram. Referring to FIG. 11, there are 12 classes whose frequency is 1 or more. The hue class is “10” and the frequency is “2500” which is the maximum. FIG. 12 is a second diagram illustrating an example of a hue histogram. The first threshold value is “750” which is 30% of the maximum frequency. FIG. 13 is a third diagram illustrating an example of a hue histogram. FIG. 13 shows a hue histogram after removing the noise hue. By removing the noise hue, the hue is limited to the 4th class hue.

  The adjacent hue integration unit 334 outputs a class having a frequency exceeding the first threshold value to the representative hue selection unit 336. In addition, when classes having a frequency exceeding the first threshold value continue, one class obtained by integrating the consecutive classes is output to the representative hue selection unit 336. Specifically, it integrates into the class with the highest frequency among consecutive classes. FIG. 14 is a fourth diagram illustrating an example of a hue histogram. FIG. 14 shows a hue histogram after the class “29” and the class “30” are integrated into the class “30”. Thereby, it is limited to the hue of the 3rd class by the integration of the hue.

  However, when the frequency class exceeding the first threshold value continues for the second threshold value or more, the adjacent hue integration unit 334 determines that the character region specified by the position information is other than characters (photograph, figure, A ruled line or the like) is output, and an error signal is output to the third attribute changing unit 335. The second threshold value is determined by the number of classes in the hue histogram, and when the hue is divided into 32 equal parts to obtain the 32nd class, good results can be obtained if the fourth class is used.

  When the error signal is input from the adjacent hue integration unit 334, the third attribute change unit 335 changes the attribute of the character region to a photo region, and displays position information indicating the position of the character region in the image data as a photo / figure region. The data is output to the encoding unit 360.

  The representative hue selection unit 336 determines the hue of the class input from the adjacent hue integration unit 334 as the hue of the character included in the character area, and outputs the input class to the character color calculation unit 338. When a plurality of classes are input from the adjacent hue integration unit 334, the hues of the plurality of classes are determined as representative hues. However, the representative hue selection unit 336 displays a photograph or a figure instead of a character in the character area specified by the position information when a class higher than the third threshold value is input from the adjacent hue integration unit 334. An area is determined, and an error signal is output to the fourth attribute changing unit 337. The third threshold value is the upper limit of the number of colors of characters included in the character area. If the third threshold value is 4, good results can be obtained.

  When an error signal is input from the representative hue selection unit 336, the fourth attribute change unit 337 changes the attribute of the character region to a photo region, and displays position information indicating the position of the character region in the image data as a photo / figure region. The data is output to the encoding unit 360.

  The character color calculation unit 338 calculates the character color from the pixel values of the color image of the plurality of pixels distributed to the class input from the representative hue selection unit 336, and outputs the calculated character color to the character region encoding unit 340. To do. Specifically, an average of RGB values of a plurality of pixels is calculated and set as a character color. When a plurality of classes are input from the representative hue selection unit 336, the character color is calculated for each of the plurality of classes.

  FIG. 15 is a functional block diagram showing detailed functions of the character area encoding unit. Referring to FIG. 15, character region encoding unit 340 receives a binary image and position information from noise removal unit 320, receives a character color from character color determination unit 330, and receives a color image. The character region encoding unit 340 includes: a character color-specific binary image generation unit 341 that generates a binary image for each character color; and a character color-specific encoding unit 343 that compresses and encodes a binary image for each character color. Including. The binary image generating unit 341 for each character color corresponds to the character color by extracting a pixel having a pixel value of the color image close to the character color from the pixels having the pixel value “1” in the binary image. A binary image for each character color is generated. Whether or not the pixel value of the color image is close to the character color may be determined by whether or not the distance in the RGB color space between the pixel value of the color image and the character color is within a predetermined range.

  For example, the character area includes seven characters “compressed by character color”, and among the seven characters in the color image, four characters “by character color” are red, one character “ni” is black, and “compressed” ”Is blue, three character colors of red, black, and blue are input to the character region encoding unit 340. The binary image generation unit 341 for each character color generates a binary image for each character color including four characters “for each character color” as a binary image corresponding to the red character color, and corresponds to the black character color. As a binary image to be generated, a binary image classified by character color including one character “ni” is generated, and as a binary image corresponding to a blue character color, binary by character color including two characters “compressed” Generate a digitized image. The character color-specific binary image generation unit 341 outputs the character color and the character color-specific binary image corresponding to the character color to the character color-specific encoding unit 343.

  The character color encoding unit 343 encodes the character color binary image using a lossless compression method to generate code data. Then, the code data, the character color, and the position information of the character region are output to the integration unit 370. Any method may be used as the lossless compression method, but here, an MMR (Modified Modified Read) compression method is used.

  Returning to FIG. 4, the photo / figure region encoding unit 360 encodes a color image other than characters (photos, diagrams, ruled lines) by the lossy compression method to generate code data. The compression method may be an irreversible compression method, and here, the JPEG (Joint Photographic Experts Group) method is used. Then, the code data is output to the integration unit 370. The integration unit 370 combines the code data input from the character region encoding unit 340, the character color and the position information of the character region, and the code data input from the photo / figure region encoding unit 360 as one file. Output. The code data is output as separate code data.

  FIG. 16 is a flowchart illustrating an example of the flow of compression processing. The compression process is a process executed by the image processing unit 30 that executes a compression program. Referring to FIG. 16, image processing unit 30 determines whether image data has been received (step S01). The process waits until image data is received (NO in step S01). If image data is received, the process proceeds to step S02. That is, the image processing is processing that is executed by receiving image data. The image data is received by the image processing unit 30 reading the image data from the RAM 112 in accordance with an instruction from the CPU 111. The image data is a color image in which one pixel includes three values of RGB (8 bits).

  Next, a brightness image is generated by performing pre-processing on the color image (step S02), and a photo / figure region extraction process for extracting a photograph region and a diagram region is performed (step S03), thereby extracting a character region. Character area extraction processing is executed (step S04). The character region extraction process will be described later, but a character region is extracted from the brightness image extracted in step S02. The character area is specified by position information indicating a position in the color image. In the character area extraction process, a plurality of character areas may be extracted.

  The processes in steps S05 to S08 are executed for one character area. For this reason, when a plurality of character regions are extracted, one of them is set as a processing target, and the processing after step S05 is executed. In step S05, the character area of the brightness image is converted into a binary image by removing noise from the character area of the brightness image generated in step S02. It is determined whether or not the character area of the brightness image has a character attribute based on whether or not noise has been removed (step S06). If it is a character attribute, the process proceeds to step S07, and if it is not a character attribute, the process proceeds to step S10.

  In step S07, a character color determination process for determining the color of the character included in the character region of the color image is executed. Then, it is determined whether or not the character area has a character attribute according to the state of the hue of the character included in the character area of the color image (step S08). If it is a character attribute, the process proceeds to step S09, and if it is not a character attribute, the process proceeds to step S10.

  In step S09, a character area encoding process for encoding the character area of the binary image is executed, and the process proceeds to step S10. At this stage, the character area of the binary image is compression encoded. In step S10, it is determined whether there is an unprocessed character area to be processed next. If such a character area exists, the character area is set as a processing target, and the process returns to step S05. If the character area does not exist, the process proceeds to step S11.

  In step S11, the photo / figure region encoding process is performed to compress and encode the photo / figure region extracted in step S03 and the region determined not to be a character in step S04, step S05, and step S07. To do. When a plurality of photograph / figure areas are extracted, compression coding is performed for each of the plurality of photograph / figure areas. For this reason, one of the plurality of photograph / drawing areas is set to be processed, and step S11 is executed. Then, it is determined whether or not there is an unprocessed photo / drawing area to be processed next (step S12). If such a photograph / character area exists, the photograph / character area is set as a processing target, and the process returns to step S11. If not, the process proceeds to step S13.

  In step S13, code data obtained by encoding the character area of the binary image in step S09, the character color of the character area, the position information of the character area, and the photograph / character area of the color image are encoded in step S11. The compressed code data is integrated to generate one compressed data.

  FIG. 17 is a flowchart illustrating an example of the flow of preprocessing. The preprocessing is processing executed in step S02 in FIG. Referring to FIG. 17, the background portion of the color image is removed (step S21). Removes the background color with low density in the color image and extends the contrast. Thereby, the tendency of a color image can be adjusted. Then, the color image from which the background is removed is converted into a brightness image (step S22). By converting the value of each pixel value (24 bits) of the color image into lightness (8 bits), the color image is converted into a lightness image. Next, the brightness image is smoothed (step S23). Since the noise is removed by smoothing the brightness image, the accuracy of the subsequent processing can be improved.

  FIG. 18 is a flowchart showing an example of the flow of the photograph / drawing area extraction process. Referring to FIG. 18, the brightness image generated in step S02 of FIG. 16 is binarized to generate a binary image (step S31). Then, the generated binary image is labeled (step S32). By labeling, a rectangle circumscribing a set of a plurality of pixels having a value of “1” in the binary image is extracted as a block. A plurality of blocks may be extracted from the binary image.

  The processes in steps S33 to S35 are processes executed for one block. For this reason, when a plurality of blocks are extracted, one of them is set as a processing target, and the processing after step S33 is executed. In step S33, the feature amount of the processing target block of the binary image generated in step S31 is calculated. Then, the attribute of the processing target block is determined from the calculated feature amount. The attributes include a character attribute, a figure attribute, a photograph attribute, and a background attribute. In step S34, if the attribute of the processing target block is a copy attribute or a graphic attribute, the process proceeds to step S35, and if not, step S35 is skipped and the process proceeds to step S36. In step S35, the processing target block is determined to be a photo / character area, and position information indicating the position of the processing target block in the binary image is stored as position information of the photo / character area. For example, the coordinates in the binary image of the two vertices opposed to the processing target block are stored in the area for storing the position information of the photo / character area in the RAM 112.

  In step S36, it is determined whether there is an unprocessed block to be processed next. If such a block exists, the block is set as a processing target, and the process returns to step S33. If not, the process returns to the compression process.

  FIG. 19 is a flowchart illustrating an example of the flow of character area extraction processing. The character area extraction process is a process executed in step S04 in FIG. The image processing unit 30 generates an edge image by filtering the lightness image generated in step S02 of FIG. 16 using, for example, a differential filter (step S41). Then, the inclination of the edge image is corrected (step S42). The edge image is rotated so that the line composed of the extracted edge pixels is horizontal or vertical. This is because image data may be input from the image reading unit 20 and the document may be read while being tilted. Then, a ruled line is detected from the edge image, and the ruled line is removed from the edge image (step S43). This is to improve the accuracy of character determination in the next step S45.

  In the next step S44, candidate blocks as character area candidates are extracted from the edge image. Specifically, first, a line space is excluded from the processing target, and edge pixels of the edge image are connected to neighboring edge pixels to extract a rectangle circumscribing a set of a plurality of connected edge pixels as a candidate block. The candidate block corresponds to a circumscribed rectangle of a character string in one line in the image. A plurality of candidate blocks may be extracted.

  The processes in steps S45 to S48 are executed for one candidate block. For this reason, when a plurality of candidate blocks are extracted, one of them is set as a processing target, and the processing after step S45 is executed. In step S45, it is determined whether the candidate block to be processed has a character attribute. The determination method is determined based on, for example, the aspect ratio of the candidate block, the length in the horizontal direction, the length in the vertical direction, the ratio of edge pixels in the edge image of the candidate block, and the like. If the candidate block to be processed is a character attribute, the process proceeds to step S46; otherwise, the process proceeds to step S47.

  In step S46, the candidate block determined as the character attribute is determined as the character area, and the position information indicating the position of the candidate block in the edge image is stored as the position information of the character area. For example, the coordinates in the edge image of the two vertices opposed by the candidate block are stored in the area for storing the position information of the character area in the RAM 112.

  On the other hand, in step S47, the attribute of the candidate block that has not been determined to be a character attribute is changed to a photo attribute, and the process proceeds to step S48. Specifically, in order to store the position information indicating the position of the candidate block in the edge image as the position information of the photograph area, the position information of the candidate block is stored in the area for storing the position information of the photograph area in the RAM 112.

  In step S48, it is determined whether there is an unprocessed candidate block to be processed next. If such a candidate block exists, the candidate block is set as a processing target, and the process returns to step S45. If not, the process returns to the compression process.

  FIG. 20 is a flowchart illustrating an example of the flow of noise removal processing. The noise removal process is a process executed in step S05 in FIG. Referring to FIG. 20, the image processing unit 30 reads the position information of the character area from the RAM 112 (step S51). Then, a brightness histogram is generated based on the pixel values of a plurality of pixels included in the character area of the brightness image generated in step S02 of FIG. 16 (step S52). Of the plurality of pixels included in the character area of the brightness image, pixels having a brightness lower than a predetermined threshold are removed as halftone noise (step S53). Specifically, the pixel value of a pixel whose brightness is lower than the threshold value is set to “0”. Further, a threshold for background removal is set between the brightness of the character and the brightness of the background from the frequency of the histogram (step S54). If the threshold value can be set, the process proceeds to step S55. If the threshold value cannot be set, the process proceeds to step S57. This is because when the threshold value for background removal cannot be set, there is a high possibility that a photo or a figure is displayed in the character area instead of characters.

  In step S55, the character area of the brightness image is binarized using a threshold value for background removal, and a binary image is generated. In the next step S56, the generated binary image is associated with the position information, and the process returns to the compression process. Specifically, the set of the binary image and position information is stored in a predetermined area of the RAM 112.

  On the other hand, in step S57, the attribute of the character area for which the threshold value could not be set is changed to a photo attribute, and the process returns to the compression process. Specifically, the position information of the character area is stored in an area for storing the position information of the photo area in the RAM 112.

  FIG. 21 is a flowchart illustrating an example of the flow of character color determination processing. The character color determination process is a process executed in step S07 in FIG. Referring to FIG. 21, the image processing unit 30 reads a pair of a binary image and position information from the RAM 112 (step S61). Thinning processing is executed to narrow the line width of the characters in the read binary image (step S62). By executing the thinning process, a binary image in which the value of the character center pixel is “1” and the values of the other pixels are “0” is generated.

  In step S63, a hue histogram of the character area is generated. The position information read in step S61 indicates the position of the character area in the color image. From this position information, the pixel in the color image corresponding to each pixel included in the binary image generated in step S62 can be specified. Therefore, a hue histogram is generated based on the pixel value of the pixel of the color image corresponding to the pixel having the value “1” in the binary image generated in step S62. In this case, the hue histogram includes a class of 32 sections obtained by dividing the hue into 32 equal parts, a white class, and a black class.

  Then, the pixel of the hue that causes noise is removed from the processing target (step S64). Specifically, in the hue histogram, a first threshold value is set based on the maximum frequency, pixels assigned to a frequency class lower than the first threshold value are regarded as noise, and the pixel is excluded from the processing target. To do. By removing noise, it is possible to improve the accuracy of character color detection. The first threshold value is preferably 30% of the maximum frequency.

  Next, it is determined whether or not three or more frequency classes exceeding the first threshold value continue in the hue histogram (step S65). If three or more classes having a frequency exceeding the first threshold value continue, the process proceeds to step S71; otherwise, the process proceeds to step S66. This is because, when three or more classes having a frequency exceeding the first threshold value continue, there is a high possibility that the character area specified by the position information represents a photograph or a figure instead of characters. In addition, although the maximum threshold value with which the frequency class exceeding the first threshold value continues is 2, this threshold value is not limited to this and may be determined according to the class of the hue histogram. it can.

  In step S66, when the frequency class exceeding the first threshold value continues, the continuous class is integrated into the class of the representative hue having the maximum frequency. Thereby, the number of character colors can be reduced and the compression rate can be improved by making pixels close to the character color into one representative hue class. Moreover, the color of the outline part and center part of a character can be made into the same color, and the appearance of a character can be improved.

  In the next step S67, it is determined whether the number of representative hues is 4 or more. If the number of representative hues is 4 or more, the process proceeds to step S71; otherwise, the process proceeds to step S68. This is because when the number of representative hues is 4 or more, the color image includes four or more colors in the character area, and there is a high possibility that a photograph or a figure is represented instead of characters. The character area is changed to the photo attribute when the number of classes of the representative hue is 4 or more. However, the threshold value “4” is not limited to this, and the character area is changed according to the class of the hue histogram. Can be determined.

  In step S68, the character color is calculated. The character color is calculated based on the representative hue class. If there are a plurality of representative hue classes, one of the plurality of representative hue classes is set as a processing target in order to calculate a character color for each of the plurality of representative hue classes. Then, in the hue histogram, the character color is calculated from the pixel values in the color image of a plurality of pixels assigned to the representative hue class to be processed. The character color is an RGB value obtained by averaging the RGB values of a plurality of pixels.

  In the next step S69, the character color, the binary image, and the position information are associated, and the process proceeds to step S70. Specifically, a set of the character color calculated in step S68, the binary image from which noise is removed in step S64, and the position information is stored in a predetermined area of the RAM 112.

  In the next step S70, it is determined whether or not there is an unprocessed representative hue class to be processed next. If there is an unprocessed representative hue class, the process returns to step S08; otherwise, the process returns to the compression process.

  On the other hand, when the process proceeds to step S71, when it is determined in step S65 that three or more classes having a frequency exceeding the first threshold value are consecutive, or in step S67, there are four or more representative hue classes. This is the case. In this case, since there is a high possibility that a photo or a figure is represented in the character area instead of a character, in step S71, the attribute of the character area is changed to a photo attribute, and the process returns to the compression process. Specifically, the position information of the character area is stored in the area for storing the position information of the photo area in the RAM 112.

  FIG. 22 is a flowchart showing an example of the flow of character area encoding processing. The character area encoding process is a process executed in step S09 in FIG. Referring to FIG. 22, the image processing unit 30 reads a set of a character color, a binary image, and position information from the RAM 112 (step S81). When a plurality of sets of the same position information are stored as a set of the character color, the binary image, and the position information, one of the sets is read out. Then, a binary image for each character color is generated (step S82). The binary image for each character color is an image in which the pixel value of a pixel of a color close to the character color is “1” and the pixel values of other pixels are “0”. Specifically, the image processing unit 30 extracts a pixel having a pixel value of a color image close to a character color from pixels having a pixel value “1” in the binary image, thereby corresponding to the character color. A binary image for each character color is generated. Whether the pixel value of the color image is close to the character color is based on the condition that the distance in the RGB color space between the pixel value of the color image and the character color is within a predetermined range.

  In the next step S83, the binary image for each character color is compression-encoded to generate code data. The compression coding here is a lossless compression method, for example, an MMR compression method. In the next step S84, the character color, the code data, and the position information are associated with each other, and the process proceeds to step S85. Specifically, the character color and position information read in step S81 and the set of code data generated in step S84 are stored in a predetermined area of the RAM 112.

  In the next step S85, it is determined whether or not there is a set of a character color, a binary image, and position information to be processed next. If such a group exists, the process returns to step S81, and if not, the process returns to the compression process.

  As described above, MFP 100 according to the present embodiment thins a character included in a character region extracted from a color image, and color image pixels corresponding to a plurality of character center pixels constituting the thinned character. The character color is determined based on the color. Then, the character area of the binary image obtained by binarizing the color image is encoded, and the code data, the determined character color, and the position of the character area in the color image are associated with each other. Since the character area of the binary image obtained by binarizing the color image is encoded, the entire character can be binarized as a character of the same color. In addition, since the character color is determined based on the color of the pixel of the image data corresponding to the plurality of character center pixels constituting the thinned character, the character color can be accurately detected. As a result, the color of characters in the image can be detected easily and with high accuracy.

  In addition, since the center pixel is extracted as the character center pixel from the continuously arranged black pixels in each of the main scanning direction and the sub-scanning direction, the character center pixel can be easily extracted, and the processing speed can be reduced. Can be fast.

  Furthermore, at least one pixel obtained by removing a predetermined number of pixels from both ends of a plurality of black pixels arranged continuously in the main scanning direction, and both ends of the plurality of black pixels arranged continuously in the sub-scanning direction Since at least one pixel excluding a predetermined number of pixels is used as the character center pixel, the character center pixel can be easily extracted, and the processing speed can be increased.

  In the above-described embodiment, the MFP 100 has been described as an example of the image processing apparatus. However, an image compression method for causing the MFP 100 to execute the compression processing illustrated in FIGS. 16 to 21 or illustrated in FIGS. Needless to say, the invention can be understood as an image compression program for causing the MFP 100 or the PCs 3, 3A, 3B, and 3C to execute the compression process.

  The embodiment disclosed this time should be considered as illustrative in all points and not restrictive. The scope of the present invention is defined by the terms of the claims, rather than the description above, and is intended to include any modifications within the scope and meaning equivalent to the terms of the claims.

<Appendix>
(1) The image processing apparatus according to claim 4, wherein the first threshold value is determined based on a maximum frequency value.
(2) The attribute changing unit changes the attribute of the character region to an attribute other than the character attribute when the hue class whose frequency exceeds the first threshold value exceeds the second threshold value. Item 10. The image processing device according to Item 8 or 9.
(3) The character color determination means further includes a selection means for selecting, as a representative hue, a hue class whose frequency exceeds a first threshold value from the plurality of hue classes.
The attribute changing unit changes the attribute of the character region to an attribute other than a character attribute when a representative hue whose frequency exceeds a first threshold exceeds a third threshold. The image processing apparatus according to 9.
(4) The image processing apparatus according to claim 1, wherein the character area encoding unit includes a reversible compression unit that reversibly compresses the binarized data.
(5) a photograph / figure area extracting means for extracting a photograph / figure attributed photograph / figure area representing a photograph or a figure from a color image;
The image processing apparatus according to claim 1, further comprising a photograph / graphic region encoding unit that encodes a color image of the photograph / graphic region.
(6) The character area encoding means includes lossless compression means for reversibly compressing the character area of the binary image,
The image processing apparatus according to (5), wherein the photograph / graphic region encoding unit includes an irreversible compression unit that performs irreversible compression on the photograph / figure region of the color image.

1 is a diagram showing an overall outline of an image forming system in one embodiment of the present invention. 1 is a perspective view showing an appearance of an MFP. 2 is a block diagram illustrating an example of a hardware configuration of an MFP. FIG. It is a functional block diagram which shows an example of the function of an image process part. It is a functional block diagram which shows the detailed function of a character area extraction part. 3 is a functional block diagram showing detailed functions of a noise removing unit 320. FIG. It is a functional block diagram which shows the detailed function of a character color determination part. It is a figure which shows an example of the binary image containing the character "A". (A) is a figure which shows an example of a main scanning direction center pixel, (B) is a figure which shows an example of a subscanning direction center pixel, (C) is a figure which shows an example of a character center pixel. It is a figure which shows an example of the class of a hue histogram. It is a 1st figure which shows an example of a hue histogram. It is a 2nd figure which shows an example of a hue histogram. It is a 3rd figure which shows an example of a hue histogram. It is a 4th figure which shows an example of a hue histogram. It is a functional block diagram which shows the detailed function of a character area encoding part. It is a flowchart which shows an example of the flow of a compression process. It is a flowchart which shows an example of the flow of pre-processing. It is a flowchart which shows an example of the flow of a photograph / figure area | region extraction process. It is a flowchart which shows an example of the flow of a character area extraction process. It is a flowchart which shows an example of the flow of a noise removal process. It is a flowchart which shows an example of the flow of a character color determination process. It is a flowchart which shows an example of the flow of a character area encoding process.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Image forming system 2 Network 3 3, 3A, 3B, 3C Computer 4 Digital camera 5 Digital video camera 6 Mobile information terminal 9 Operation panel 20 Image reading part 20 Image reading part 30 Image processing part 40 image forming unit, 50 paper feeding unit, 60 facsimile unit, 61 communication control unit, 100 MFP, 101 main circuit, 111 CPU, 112 RAM, 113 ROM, 114 display unit, 115 operation unit, 116 HDD, 117 data communication control , 118 LAN terminal, 119 serial communication interface terminal, 119A memory card, 301 preprocessing unit, 310 character area extraction unit, 311 edge image generation unit, 312 tilt correction unit, 313 ruled line removal unit, 314 candidate block extraction unit, 315 Judgment part, 316 sentences Area determination unit, 317 first attribute change unit, 320 noise removal unit, 321 brightness histogram generation unit, 323 halftone dot / background removal unit, 325 second attribute change unit, 330 character color determination unit, 331 thinning unit, 332 hue Histogram generation section, 333 noise hue removal section, 334 adjacent hue integration section, 335 third attribute change section, 336 representative hue selection section, 337 fourth attribute change section, 338 character color calculation section, 340 character area encoding section, 341 Value image generation unit, 343 Character color coding unit, 350 photo / figure region extraction unit, 360 photo image coding unit, 370 integration unit.

Claims (13)

A character area extracting means for extracting a character area having a character attribute representing a character from a color image;
Thinning means for thinning characters included in the character region;
A character color determining means for determining a character color based on a color of a pixel of the color image corresponding to a plurality of character center pixels constituting the thinned character;
Binarization means for converting the color image into a binary image by binarizing;
A character area encoding means for encoding the character area of the binary image and generating code data;
An image processing apparatus comprising: association means for associating the generated code data, the determined character color, and a position of the character region in the color image. The thinning unit is a main scanning direction detecting unit that extracts a central pixel from black pixels continuously arranged in the main scanning direction in the binary image;
Sub-scanning direction detecting means for extracting a central pixel from black pixels continuously arranged in the sub-scanning direction intersecting with the main scanning direction;
The image processing apparatus according to claim 1, further comprising: a determining unit that determines the center pixel extracted by the main scanning direction detection unit and the center pixel extracted by the sub-scanning direction detection unit as the character center pixel. . The thinning means extracts main scanning direction detection means for extracting at least one pixel excluding a predetermined number of pixels from both ends of a plurality of black pixels arranged continuously in the main scanning direction as a first candidate pixel;
Sub-scanning direction detecting means for extracting at least one pixel excluding a predetermined number of pixels from both ends of a plurality of black pixels continuously arranged in the sub-scanning direction intersecting with the main scanning direction;
The image processing apparatus according to claim 1, further comprising: a determination unit that determines a pixel that is both the first candidate pixel and the second candidate pixel as the character center pixel. The character color determining means is a classification means for classifying each of the plurality of character center pixels into one of a plurality of hue classes based on a color before thinning of each of the plurality of character center pixels.
The image processing apparatus according to claim 1, further comprising a noise removing unit that removes, as noise, pixels classified into a hue class whose frequency is equal to or less than a first threshold value. The character color determining means includes a classification means for classifying each of the plurality of character center pixels into one of a plurality of hue classes based on a color before thinning of each of the plurality of character center pixels; Selection means for selecting, as a representative hue, a hue class whose frequency exceeds a first threshold value from among the classes,
The image processing device according to claim 1, wherein the character color is determined based on a color before thinning of the character center pixel classified into a hue class of the selected representative hue among the plurality of character center pixels. .   The selection means includes hue integration means for selecting, as a representative hue, a hue class having the highest frequency among consecutive hue classes when two or more hue classes having a frequency exceeding the first threshold value are consecutive. 5. The image processing apparatus according to 5. The character color determining means determines a plurality of the character colors respectively corresponding to the plurality of representative hues when a plurality of the representative hues are selected by the selecting means;
The character region encoding means includes, in each of the plurality of representative hues, a color of a pixel of the color image corresponding to the black pixel among the black pixels included in the character region of the binary image. Binarized data generation unit for each character color that generates a binary image for each character color corresponding to the representative hue by extracting pixels that approximate the corresponding character color;
A character color coding unit that codes a plurality of character color binarized data respectively corresponding to the determined plurality of character colors, and generates character color code data.
The association means associates, for each of the determined plurality of character colors, the character color, the character color code data corresponding to the character color, and the position of the character region in the image data. The image processing apparatus according to claim 5, comprising color-specific association means.   The character color determining means includes a classification means for classifying each of the plurality of character center pixels into one of a plurality of hue classes based on a color before thinning of each of the plurality of character center pixels; The image processing apparatus according to claim 1, further comprising: an attribute changing unit that changes an attribute of the character region to an attribute other than the character attribute based on a class frequency distribution. A character area extracting means for extracting a character area having a character attribute representing a character from a color image;
A photograph / figure region extracting means for extracting a photograph / figure attribute / photo / figure region representing a photograph or a figure from a color image;
Thinning means for thinning characters included in the character region;
Classifying means for classifying each of the plurality of character center pixels into one of a plurality of hue classes based on the color before thinning of each of the plurality of character center pixels constituting the thinned character;
An image processing apparatus comprising: an attribute changing unit that changes an attribute of the character region to an attribute other than the character attribute based on the frequency distribution of the plurality of hue classes. Extracting a character region having a character attribute representing a character from a color image;
Thinning characters included in the character region;
Determining a character color based on a color of a pixel of the color image corresponding to a plurality of character center pixels constituting the thinned character;
Converting the color image into a binary image by binarizing the color image;
Encoding the character region of the binary image to generate code data;
An image processing method comprising: associating the generated code data, the determined character color, and a position of the character region in the color image. Extracting a character region having a character attribute representing a character from a color image;
Extracting a photo / picture attribute of a picture / picture attribute representing a picture or figure from a color image;
Thinning characters included in the character region;
Classifying each of the plurality of character center pixels into one of a plurality of hue classes based on the color before thinning of each of the plurality of character center pixels constituting the thinned character;
Changing the attribute of the character region to an attribute other than the character attribute based on the frequency distribution of the plurality of hue classes. Extracting a character region having a character attribute representing a character from a color image;
Thinning characters included in the character region;
Determining a character color based on a color of a pixel of the color image corresponding to a plurality of character center pixels constituting the thinned character;
Converting the color image into a binary image by binarizing the color image;
Encoding the character region of the binary image to generate code data;
An image processing program that causes a computer to execute the step of associating the generated code data, the determined character color, and the position of the character region in the color image. Extracting a character region having a character attribute representing a character from a color image;
Extracting a photo / picture attribute of a picture / picture attribute representing a picture or figure from a color image;
Thinning characters included in the character region;
Classifying each of the plurality of character center pixels into one of a plurality of hue classes based on the color before thinning of each of the plurality of character center pixels constituting the thinned character;
An image processing program for causing a computer to execute the step of changing the attribute of the character region to an attribute other than the character attribute based on the frequency distribution of the plurality of hue classes.

Images