JP2015065596A - Image processing apparatus, image processing method and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve such a problem that since readability is changed when performing resolution reduction in accordance with a sort of a character, it is difficult to perform the resolution reduction by selecting suitable resolution.SOLUTION: An image processing apparatus for generating compression image from an input image includes: generation means for generating one or more low-resolution images having resolution different from that of the input image; creation means for creating binary images for at least two of the input image and the one or more low-resolution images; selection means for selecting a binary image from the plurality of binary images on the basis of a difference of feature values between the plurality of binary images created by the creation means; and compression means for generating compression data for the input image by compressing the selected binary image.

Description

  The present invention relates to an image processing device, an image processing method, and a program.

  In recent years, with the widespread use of color printers and color scanners, there are increasing opportunities to handle image data of colored documents. However, since full-color image data has a heavy load on a storage device and a line, it is necessary to reduce the data amount by some method.

  In Patent Document 1, color images are subjected to color reduction processing, and the color information obtained from the color reduction processing and the color index image are integrated and compressed, so that compression efficiency is improved and compression with good reproducibility is performed. A method has been proposed.

JP 2003-309727 A

  In Patent Document 1, since color images with various resolutions can be input, for example, a color image having a high resolution may be input. In such a case, in order to reduce the data size, the user can specify the output resolution to a lower resolution (for example, 100 dpi), and can perform color reduction processing and image encoding at the specified resolution. is there.

  However, when a low resolution is specified, there is a problem that characters included in the image are crushed, resulting in poor readability. Further, even if the resolution is the same, there is a problem that it is difficult to convert to an appropriate low resolution because the readability of the character at the time of conversion changes depending on the character characteristics such as the complexity of the character.

  In order to solve the above problems, the present invention has the following configuration. That is, an image processing apparatus that generates compressed data from an input image, the generation unit generating one or more low-resolution images having a resolution different from the input image using the input image, and the input image And a creation unit that creates a binary image for at least two of the one or a plurality of low-resolution images, and a difference in feature amount between the plurality of binary images created by the creation unit, Selecting means for selecting a binary image from a plurality of binary images; and compression means for generating compressed data for the input image by compressing the selected binary image.

  When the resolution of image data including characters is reduced, conversion to a resolution corresponding to the characteristics of each character can be performed, and the readability of each character can be maintained.

1 is a block diagram showing a configuration example of an image processing system according to the present invention. FIG. 2 is a block diagram showing an example of the hardware configuration of an MFP. The block diagram which shows the software structural example of a data processing part. The flowchart of the whole process which concerns on 1st Embodiment. The figure which shows the example of the input data which concerns on 1st Embodiment. The figure for demonstrating the resolution and the readability of the character with respect to a font. The figure which shows the calculation result of the closed loop feature-value which concerns on 1st Embodiment. The figure for demonstrating the closed loop feature-value which concerns on 1st Embodiment. The figure for demonstrating the color reduction process which concerns on 1st Embodiment. The figure for demonstrating the structure of the compressed data which concerns on 1st Embodiment. The schematic diagram of the PDF structure concerning a 1st embodiment. The figure for demonstrating the closed loop feature-value which concerns on 2nd Embodiment. The figure for demonstrating the closed loop feature-value which concerns on 2nd Embodiment. The figure which shows the example of the input data which concerns on 3rd Embodiment. The schematic diagram of the PDF structure concerning a 3rd embodiment.

[Detailed description of the issue]
The problem handled by the present invention will be described in detail with reference to FIG.

  FIG. 6 is a diagram for explaining the readability of characters with respect to resolution and font. Bitmap data 601 to 609 respectively represent binary image bitmap data. The horizontal axis represents the resolution (300, 200, 100 dpi), and the vertical axis represents the font type (MS Gothic, HGP Creative English Gothic UB), and the type of Kanji and Hiragana. The character sizes (font sizes) are all the same, for example, a character size of 8 pt.

  First, the bitmap data 601 to 603 indicate a character string of “Kanji” of MS Gothic. Here, the readability of the bitmap data 601 and 602 having resolutions of 300 dpi and 200 dpi is maintained. On the other hand, in the bitmap data 603 with a resolution of 100 dpi, particularly “Den” and “True” characters are crushed and the readability is poor.

  Next, the bitmap data 604 to 606 indicate a character string of “Hiragana” of MS Gothic. Here, in the case of MS Gothic “Hiragana”, the bitmap data 606 having a resolution of 100 dpi maintains readability because the characters are not significantly collapsed. That is, even in the same font, characters may be crushed at a resolution of 100 dpi as shown in the bitmap data 603, or readability may be maintained even at a resolution of 100 dpi as shown in the bitmap data 606. . In general, at a low resolution, “Kanji” including more complicated characters tends to be crushed compared to “Hiragana” and “Alphabet”.

  Bitmap data 607 to 609 indicate “Kanji” of HGP Creative English Gothic UB. Here, in the case of “Kanji” of HGP Creative English Gothic UB, the readability of the bitmap data 607 having a resolution of 300 dpi is maintained. On the other hand, the bitmap data 608 having a resolution of 200 dpi and the bitmap data 609 having a resolution of 100 dpi are distorted as a whole, resulting in poor readability. In other words, due to the difference between the fonts of MS Gothic and HGP Creative Horn Gothic UB, characters may be crushed at a resolution of 100 dpi as shown in bitmap data 603, or characters at a resolution of 200 dpi as shown in bitmap data 608. May collapse.

  As described above, when converting the resolution, the readability of the character changes depending on the resolution, the type of font, and the type of simple character (for example, hiragana) or complex character (for example, kanji). Therefore, there is a problem that it is difficult to convert each character to an appropriate resolution (low resolution).

  Embodiments according to the present invention for solving the above-described problems will be described below.

<First Embodiment>
[System configuration]
Hereinafter, embodiments for carrying out the present invention will be described with reference to the drawings. FIG. 1 is a block diagram illustrating a configuration example of an image processing system according to the present embodiment.

  In FIG. 1, an MFP (Multi-Function Peripheral) 100, which is a multifunction machine that realizes a plurality of types of functions (copying function, printing function, transmission function, etc.), is connected to the LAN 102 constructed in the office A. . The LAN 102 is also connected to the network 104 via the proxy server 103 in order to communicate with an external device.

  The client PC 101 receives transmission data from the MFP 100 via the LAN 102 and uses functions that the MFP 100 has. For example, the client PC 101 can also print the printed matter based on the print data by the MFP 100 by transmitting the print data to the MFP 100. The configuration in FIG. 1 is an example, and a plurality of offices having the same components as the office A may be connected on the network 104.

  The network 104 is a communication network typically realized by the Internet, LAN, WAN, telephone line, dedicated digital line, ATM, frame relay line, communication satellite line, cable TV line, data broadcasting radio line, and the like. is there. Note that any configuration may be used as the network 104 as long as data can be transmitted and received. Each of the various terminals of the client PC 101 and the proxy server 103 has standard components mounted on a general-purpose computer. For example, a CPU, RAM, ROM, hard disk, external storage device, network interface, display, keyboard, mouse, and the like.

  The system configuration shown above is an example, and other devices may be included in the office A. In FIG. 1, each device shows only one device, but a plurality of devices may be included.

[Hardware configuration]
FIG. 2 is a diagram showing a detailed hardware configuration example of the MFP according to the embodiment of the present invention.

  The MFP 100 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, and an operation unit 203 that is a user interface. The control unit 204 is a controller that is connected to the scanner unit 201, the printer unit 202, and the operation unit 203, and on the other hand, is connected to the LAN 209 to input and output image information and device information. A CPU 205 is a controller that controls the entire system. A RAM 206 is a system work memory for the CPU 205 to operate, and is also an image memory for temporarily storing image data. A ROM 210 is a boot ROM, and stores programs such as a system boot program. The storage unit 211 is a hard disk drive, and stores system control software and image data.

  An operation unit I / F 207 is an interface unit with the operation unit 203, and outputs image data to be displayed on the operation unit 203 to the operation unit 203. An operation unit I / F 207 serves to transmit information input by the user of the image processing apparatus from the operation unit 203 to the CPU 205. A network I / F 208 connects the image processing apparatus to the LAN 209 and inputs / outputs packet format information. The above devices are arranged on the system bus 216. The image bus 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 configured by, for example, a PCI bus or IEEE1394. The following devices are arranged on the image bus 217. A RIP (Raster Image Processor) 213 implements a so-called rendering process in which a PDL (Page Description Language) code is analyzed and developed into a bitmap image having a specified resolution. The device I / F 214 connects the scanner unit 201 that is an image input device to the control unit 204 via the signal line 218. The device I / F 214 connects the printer unit 202 as an image output device to the control unit 204 via the signal line 219, and performs synchronous / asynchronous conversion of image data. The data processing unit 215 will be described later.

[Description of data processing unit]
Next, the data processing unit 215 in FIG. 2 will be described with reference to FIG. The data processing unit 215 includes a resolution conversion unit 301, a color reduction processing unit 302, a color information sorting unit 303, a same color integration unit 304, a binary image creation unit 306, a closed loop feature quantity calculation unit 307, a binary image selection unit 308, a binary value. An image compression unit 309, a data combination unit 310, and a format conversion unit 311 are included. The data processing unit 215 receives the input data 300 and outputs output data 305. The input data 300 is bitmap data read from the scanner unit 201 or bitmap data stored in the storage unit 211. The output data 305 is electronic document data, and processing such as output by the printer unit 202, storage in the storage unit 211, and transmission to an external device connected to the network via the LAN 209 is performed. Here, the electronic document data is an electronic document format such as PDF (Portable Document Format), XPS (XML Paper Specification), OfficeOpenXML, and the like. In this embodiment, the input data 300 is bitmap data obtained by reading a paper document with the scanner unit 201, and the resolution is 300 dpi.

  The resolution conversion unit 301 converts the input data 300 to a predetermined resolution. In this embodiment, the resolution conversion unit 301 mainly performs resolution conversion to a low resolution lower than the resolution of the input data 300. For example, resolution conversion from 300 dpi to 200 dpi and resolution conversion from 300 dpi to 100 dpi are performed. Do. Although details will be described later, a closed-loop feature amount is calculated for each of the input data 300, the 200 dpi image after resolution conversion, and the 100 dpi image.

  The color reduction processing unit 302 performs color reduction processing on the input data 300 to a predetermined number of colors. In this color reduction process, it means that the data of 8 bits (8-8-8) for each of the three colors of full color RGB is reduced to the number of bits such as 2-2-2, 3-2-2, 3-3-3 bits, etc. To do. The method of selecting the number of bits is selected depending on the accuracy with which the color determination is desired. Since a known technique may be used for the color reduction process, in this embodiment, the color reduction process will be briefly described with reference to FIG.

  FIG. 9 is a schematic diagram showing how an index color image and color information for each color are generated from the input data 300 by the color reduction process. FIG. 9A shows an example of the input data 300, which includes a black character region 901, a red character region 902, and a white background. FIG. 9B shows a result of performing the color reduction processing on FIG. 9A, and index color images 903 to 905 are generated. Here, when the input data 300 is data obtained by reading a paper document with the scanner unit 201, color variation on the paper document occurs. For example, in the case of the character region 902, a plurality of similar red colors are generated as a result of the color reduction process, and therefore, a plurality of index color images are generated instead of the one shown in the index color image 904.

  Further, each index color image is associated with color information as shown in FIG. Here, the color information includes data indicating color values (color centroid), position coordinates (distribution range), and the number of pixels. In the case of a character area, the color value (color centroid) indicates the color of a pixel constituting the character. The position information (distribution range) is, for example, data including data of the coordinates of the upper left corner and the coordinates of the lower right corner of the index color image in the input data 300. In this way, an index color image and color information for each color are generated from the input data 300.

  The color information sorting unit 303 sorts the above color information according to the number of pixels. As a result of the sorting, the index color image is basically positioned higher as the number of pixels is larger. Here, an index color image having the color value (color centroid) of the most significant color information is set as a background color image.

  The same color integration unit 304 compares the color information sorted by the color information sorting unit 303, determines that similar colors are the same color, performs integration processing, and updates the color information subjected to integration processing. Specifically, the color values (color centroids) of the color information are compared, and if the values are relatively close (for example, the absolute value of the difference between the values is equal to or less than a predetermined value), the values are compared as the same color. Two color information is integrated. This is a process of returning the color separated into a plurality of colors to one because one of the RGB values happens to be close to the threshold value of the first subtractive color process even though the color before scanning is the same color.

  Along with the integration process, the number of pixels, color value (color centroid), and distribution range included in the color information are recalculated. Specifically, the number of pixels after integration can be obtained by adding the number of pixels included in each color information to be integrated. Further, by adding the color values (color centroids) included in each color information to be integrated and dividing by 2, it is possible to obtain the color centroid (average color) after integration. In addition, in the data of the distribution range included in each color information to be integrated, the distribution range after integration can be obtained by adopting the coordinate indicating the position of the upper left corner and the coordinate indicating the position of the lower right corner. it can. Through the above processing, the color information can be updated along with the integration of the color information. The color information update process is an example, and the present invention is not limited to this.

  The binary image creation unit 306 creates a binary image using an index color image for each color excluding an index color image having a background color.

  The closed loop feature quantity calculation unit 307 detects a closed loop from the binary image created by the binary image creation unit 306 using known contour tracking, and calculates the detected closed loop feature quantity (closed loop feature quantity).

  Here, the closed loop feature quantity as a point of the present invention will be described with reference to FIG. First, “closed loop” means a white pixel region surrounded by black pixels or a black pixel region surrounded by white pixels. In this embodiment, a white pixel region surrounded by black pixels is referred to as “closed loop (white)”, and a black pixel region surrounded by white pixels is referred to as “closed loop (black)”. For example, FIG. 8A shows a Chinese character “Den” in a binary image, and four areas indicated by an area 801 are closed loop (white). Similarly, FIG. 8B shows the hiragana character “nu” in the binary image, and three areas indicated by an area 802 are closed loop (white). FIG. 8C shows the reverse character “Den” of the Chinese character in the binary image, and the four areas indicated by the area 803 are closed loop (black). The contour tracking processing method for detecting the closed loop may use a known technique, and a detailed description thereof is omitted here.

  Next, the number of closed loops included in one character will be described as an example of the closed loop feature value. The character “den” in FIG. 8A has four closed loops (white), so the number of closed loops (white) = 4, and the character “nu” in FIG. 8B is closed loop (white). Since there are three, the number of closed loops (white) = 3. 8C has four closed loops (black), the number of closed loops (black) = 4.

  Note that the closed loop feature amount is not limited to the number of closed loops included in one character, and may be the total number of closed loops of a plurality of characters included in a character area. Further, it may be handled in units of pages and may be the total number of closed loops of all characters in the page. Further, coordinate information 805, rectangle information 806 circumscribing the closed loop, short vector information and vector information (not shown) may be used as shown in a closed loop (white) 804 in FIG. 8D. In the present embodiment, the total number of closed loops of a plurality of characters included in the character area is used.

  The binary image selection unit 308 determines the resolution at which the character readability is deteriorated based on the closed-loop feature value calculated by the closed-loop feature value calculation unit 307, and selects a binary image having a resolution higher than the determined resolution. . Details of the determination method will be described later.

  A binary image compression unit 309 performs compression on the binary image selected by the binary image selection unit 308. Color information is added to each binary image compressed data.

  The data combining unit 310 combines the background color data and the binary image compressed data to create a compressed image.

  The format conversion unit 311 converts the compressed image combined by the data combining unit 310 into an electronic document format such as PDF, XPS, or OfficeOpenXML.

[Processing flow]
Next, image processing according to the present embodiment will be described with reference to the flowcharts of FIGS. Note that a program for realizing the flowchart is stored in the ROM 210 or the storage unit 211 in FIG. 2 and executed by the CPU 205. The CPU 205 can exchange data with the data processing unit 215 using the image bus I / F 212, the system bus 216, and the image bus 217.

  A specific example of the input data 300 used for explanation in the present embodiment is shown in FIG. The input data 300 includes character areas 501 to 503, and the font and character color in each character area are as shown in FIG. The resolution of the input data 300 is 300 dpi. For the sake of comprehensive explanation, an example is given in which a plurality of types of fonts and a plurality of types of character colors are included in one page. However, the present invention is not limited to this. , And a single character color may be included.

  When the input data 300 is input to the data processing unit 215, in step S401, the resolution conversion unit 301 determines whether or not a closed-loop feature amount at the input resolution (here, 300 dpi) of the input data 300 has already been calculated. If not calculated (NO in S401), the process proceeds to S403, and if calculated (YES in S401), the process proceeds to S402. Note that the calculation of the closed-loop feature amount at the input resolution is performed in S409 described later.

  In step S403, the color reduction processing unit 302 performs color reduction processing on the input data 300 to a predetermined number of colors, and generates an index color image and color information for each color. In step S404, the color information sorting unit 303 receives the color information obtained by the color reduction processing unit 302 and sorts each color information according to the number of pixels. In step S405, the same color integration unit 304 compares the color information sorted by the color information sorting unit 303, determines that similar colors are the same color, performs integration processing, and updates the color information subjected to integration processing. .

  In S406, the same color integration unit 304 converts the color values (color centroids) included in all color information including the color information updated in S405 into luminance color difference values. Further, the same color integration unit 304 compares the luminance color difference values after conversion, and performs a combining process on color information having similar values (for example, color information in which the absolute value of the difference between each value is a predetermined value or less). Do. Thereby, the number of colors located in the halftone is reduced. This process corresponds to the fact that many gradations from white to black occur at the boundary between the white of the background and the black of the character part when the original document image is a black and white document, This is done to remove these gradations. Gray near white is white and gray near black is black. The other colors are processed in the same manner.

  In step S407, the same color integration unit 304 outputs the color value (color centroid) of the most significant color information among the sorted color information as a background color. In step S <b> 408, the binary image creation unit 306 creates a binary image for each color using color information other than the highest level and the index color image. The binary image created here has a size corresponding to the color distribution range held in the color information, and if it exists only in a part of the document, only that part is created.

  In step S <b> 409, the closed loop feature quantity calculation unit 307 calculates a closed loop feature quantity from the binary image created by the binary image creation unit 306. In the case of the input data in FIG. 5, closed loop feature values are calculated for each of the character regions 501 to 503. The calculation result of the closed loop feature value will be described later.

  In S410, the resolution conversion unit 301 determines whether or not the closed-loop feature amount calculation has been performed for all resolutions. If closed-loop feature value calculation has been completed for all resolutions (YES in S410), the process proceeds to S411. If closed-loop feature values have not been calculated for all resolutions (NO in S410), the process returns to S401, and the resolution conversion unit 301 determines again whether closed-loop feature values for the input resolution have already been calculated. If the closed-loop feature value at the input resolution has already been calculated (YES in S401), the process proceeds to S402.

  In S <b> 402, the resolution conversion unit 301 performs resolution conversion of the input data 300 to a low resolution. In this embodiment, resolution conversion from 300 dpi to 200 dpi and resolution conversion from 300 dpi to 100 dpi are performed. First, resolution conversion from 300 dpi to 200 dpi is performed in S402, and a closed loop feature amount at 200 dpi is calculated in S403 to S409. Thereafter, the resolution conversion from 300 dpi to 100 dpi is performed again in S402, and the closed loop feature amount at 100 dpi is calculated in S403 to S409. In this way, S401 to S410 are repeated until closed-loop feature values at all resolutions are calculated. All resolutions in the present embodiment mean three input resolutions of 300 dpi and 200 dpi and 100 dpi after resolution conversion.

  In step S411, the binary image selection unit 308 determines a resolution at which character readability deteriorates based on the closed-loop feature amount calculated by the closed-loop feature amount calculation unit 307, and a binary having a resolution higher than the determined resolution. Select an image. Here, FIG. 7 shows the calculation result of the closed loop feature quantity at each resolution in the case of the input data of FIG. The total values 701 to 703 shown in FIG. 7 are the total values of the number of closed loops (white) in the character region 501 at 300 to 100 dpi. Here, the total value 701 and the total value 702 are the total number of closed loops (white) = 7, whereas the total value 703 is changed to the total value = 4. ) In the total value (difference) is 7-4 = 3.

  In this embodiment, if the amount of change in the total value of the number of closed loops (white) is larger than a predetermined threshold, it is predicted that the character is crushed. Therefore, for example, when a predetermined threshold value = 2, since the amount of change in the total value of the closed loop (white) = 3> threshold = 2, in the character region 501, when the resolution is 100 dpi, character readability Is determined to be worse. Next, the total value 704 and the total value 705 are the total values of the number of closed loops (white) at 300 dpi and 100 dpi in the character area 502. Here, the total value 704 and the total value 705 are not changed from the total value of the number of closed loops (white) = 4. Therefore, since the change amount of the total value of the number of closed loops (white) = 0, it is determined that the character readability is maintained even in the case of 100 dpi in the character region 502. If the resolution is 200 dpi, it will be omitted for the sake of simplicity.

  The total value 706 and the total value 707 are the total values of the numbers of closed loops (white) at 300 dpi and 200 dpi in the character area 503. Here, the total value 706 is the total value of the number of closed loops (white) = 7, whereas the total value 707 is changed to the total value = 4. The amount of change is 7-4 = 3. For example, if a predetermined threshold value = 2, the amount of change in the total value of the closed loop (white) = 3> threshold = 2, and therefore, in the character region 503, the character readability deteriorates in the case of 200 dpi. It is determined that If the resolution is 100 dpi, it will be omitted for the sake of simplicity. Although explanation is omitted, in order to improve the determination accuracy, information such as known font recognition and font size may be used in combination.

  As described above, since the resolution at which the readability of the characters in the character regions 501 to 503 is deteriorated is determined, the binary image selection unit 308 selects a binary image having a resolution higher than the resolution determined to be crushed. . Here, a binary image that is larger than the resolution determined to be crushed and has the minimum resolution is selected from the plurality of resolution images. That is, a binary image having a resolution of 200 dpi is selected for the character area 501, 100 dpi for the character area 502, and 300 dpi for the character area 503. Note that the accuracy of the threshold for the change amount may be changed according to, for example, a character unit, a character string unit, or a page unit.

  In step S412, the binary image compression unit 309 compresses the binary image selected by the binary image selection unit 308 by a method such as MMR. The MMR compressed data compressed by MMR and the color information are collectively referred to as “color compressed data”. The configuration of the color compressed data will be described later.

  In S413, the data combination unit 310 combines the background color data and the color compressed data generated in S412 to generate compressed data. FIG. 10 shows a configuration example of the compressed data. First, information such as the size, background color value, and resolution of the document image input to the header portion 1001 is stored. Since the color having the largest number of pixels is basically selected as the background color, for example, when the original is printed on a color paper such as red, a red value is entered. Since it is considered that the background is often white, white determination for the background color may be performed, and if it is determined white, the background color value may be omitted. In the white determination here, for example, when each value of RGB is a certain value or more and the difference between the values is within a certain value, the background color is regarded as white.

  The header portion 1001 is followed by color compressed data 1002 for each color. Each of the color compressed data 1002 is composed of color information and MMR compressed data as described above. When the number of remaining colors excluding the background color is N, data of the same structure exists for the number of colors. Of course, if the input image is a single color original such as a blank sheet, the data of this portion is not created. In the case of a black and white document, the number of color compressed data is 1, which is almost equivalent to a binary image. If the black pixel is only a part of the original, the MMR compressed data is compressed only that part, and therefore becomes smaller than the normal MMR compressed data.

  As a method for decoding compressed data into an original image, for example, the entire area of the document is drawn with the background color stored in the header portion 1001 shown in FIG. 10, and the MMR compressed data included in the color compressed data 1002 is displayed. Decompress in the order stored. Then, the original image is decoded by overwriting according to the stored position and color using the image as a mask.

  In step S414, the format conversion unit 311 converts the compressed data generated in step S413 into an electronic document format such as PDF, XPS, or OfficeOpenXML. In the case of the input data of FIG. 5, for example, a schematic diagram of the PDF structure when converted to PDF is shown in FIG. In the data 1101, black is added as color information to a binary image generated with a resolution of 200 dpi. In data 1102, blue is added as color information to a binary image generated with a resolution of 100 dpi. In data 1103, red is added as color information to a binary image generated with a resolution of 300 dpi. Data 1104 is a background color, and white is given as color information.

  In the present embodiment, when a plurality of low-resolution images are created and each closed-loop feature amount is calculated, resolution conversion to low resolution and closed-loop feature amount calculation are performed in order. However, the present invention is not particularly limited to this, and a plurality of low-resolution images may be generated at the same time, and closed-loop feature value calculation may be performed by parallel processing.

  In this embodiment, a closed-loop feature using a total of three binary images including a binary image created from an input image (300 dpi) and a binary image created from two different low-resolution images (200 dpi, 100 dpi). The explanation for calculating the amount was given. However, the binary image created from the input image may not be used, and the closed-loop feature value calculation may be performed using the binary image created from two or more low-resolution images.

  As described above, in the case of reducing the resolution to reduce the data size, conversion to an appropriate resolution can be performed according to the characteristics of each character, so that it is possible to maintain readability even with easily crushed characters. .

<Second Embodiment>
In the first embodiment, the total number of closed loops (white) of one or more characters included in the character region is used as the feature amount of the closed loop, and the change amount of the total value is compared with a predetermined threshold value. If it is larger, it is predicted that the character is crushed. However, even if there is no change in the total value of the closed loop (white), the character may be crushed. Therefore, in this embodiment, the character is crushed using a feature amount other than the total number of closed loops. The prediction method will be described. Note that description of parts that perform the same processing as in the first embodiment is omitted.

  FIG. 12 is a schematic diagram for illustrating an example in which characters are crushed even when there is no change in the total value of the closed loop (white). First, the binary image 1201 is a binary image having a resolution of 300 dpi, and the binary image 1202 is a binary image having a resolution of 100 dpi. In any case, since the number of closed loops (white) = 4, the amount of change = 0, but the binary image 1202 is crushed. Next, the binary image 1203 is a binary image of inverted characters with a resolution of 300 dpi, and the binary image 1204 is a binary image of inverted characters with a resolution of 100 dpi. In any case, since the number of closed loops (white) = 4, the amount of change = 0, but the binary image 1204 is crushed.

  First, in the case of a binary image 1202, as shown in FIG. 13, the collapse of characters can be predicted from the amount of change in the width and height of the circumscribed rectangle of the closed loop.

  A rectangle 1301 indicates the width W1 and the height H1 of the rectangle circumscribing the closed loop at the lower left of the closed loop in the binary image 1201. A rectangle 1302 indicates a width W2 and a height H2 of the rectangle circumscribing the closed loop at the lower left of the closed loop in the binary image 1202. Here, if the amount of change in the width of the circumscribed rectangle of the closed loop (W1-W2) or the amount of change in height (H1-H2) is larger than a predetermined threshold, it is predicted that the character is crushed, If the resolution is 100 dpi, it is determined that the readability of the character is poor.

  Further, in the case of a binary image 1204, the collapse of characters is predicted using the total number of closed loops (black). Although not shown, the binary image 1203 has the number of closed loops (black) = 4, whereas the binary image 1203 has the number of closed loops (black) = 2. Therefore, if the amount of change in the total value of the closed loop (black) is larger than a predetermined threshold value, it is predicted that the character is crushed, and if the resolution is 100 dpi, it is determined that the character readability is deteriorated. To do.

  As described above, in addition to the total value of the closed loop (white) used in the first embodiment, the amount of change in the width and height of the circumscribed rectangle of the closed loop and the amount of change in the total value of the closed loop (black) are used in combination. Can judge the crushing of characters more accurately.

<Third Embodiment>
In the first embodiment, taking the input data shown in FIG. 5 as an example, the total number of closed loops of a plurality of characters included in the character area is used, and conversion to an appropriate resolution is possible in character area units. Met. However, the data size is not limited to a character area unit, and if data is converted to an appropriate resolution in character units, the data size can be further reduced. In the present embodiment, a case where conversion to an appropriate resolution is performed for each character will be described. Note that a description of the same processing as in the first embodiment will be omitted.

  FIG. 14 is an example of input data according to the present embodiment. The input data includes a character area 1401, and the font and character color of the character area 1401 are as shown in FIG. Character regions 1402 to 1405 indicate character unit regions.

  Using the example of input data shown in FIG. 14, if the same processing as in the flowchart (FIG. 4) of the first embodiment is performed in units of characters, the characters “Den” and “True” are collapsed at 100 dpi. It is predicted that On the other hand, the characters “child” and “copy” are determined to maintain the readability of the characters even in the case of 100 dpi. It should be noted that the information on the character unit area is obtained by performing a known character separation process after the binary image is created in S408.

  In the case of the input data of FIG. 14, for example, a schematic diagram of the PDF structure when converted to PDF is shown in FIG. In data 1501, black is added as color information to a binary image generated at a resolution of 200 dpi. In the data 1502, black is added as color information to a binary image generated with a resolution of 100 dpi.

  As described above, when the conversion to the resolution is performed in units of character areas, the resolution of 200 dpi is required for all the characters. On the other hand, the conversion to the resolution in units of characters makes it possible for some characters even in the case of 100 dpi. It is determined that the readability of the characters is maintained. Therefore, the data size can be further reduced.

<Other embodiments>
The present invention can also be realized by executing the following processing. That is, software (program) that realizes the functions of the above-described embodiments is supplied to a system or apparatus via a network or various storage media, and a computer (or CPU, MPU, or the like) of the system or apparatus reads the program. It is a process to be executed.

Claims (8)

An image processing device that generates compressed data from an input image,
Generating means for generating one or a plurality of low-resolution images having a resolution different from that of the input image using the input image;
Creating means for creating a binary image for at least two of the input image and the one or more low resolution images;
A selection unit that selects a binary image from the plurality of binary images based on a difference in feature amount between the plurality of binary images created by the creation unit;
An image processing apparatus, comprising: a compression unit that generates compression data for the input image by performing compression on the selected binary image. Further comprising a calculation means for detecting a pixel area surrounded by pixels constituting a character from each of the plurality of binary images created by the creation means, and calculating a feature amount of the pixel area;
The image processing apparatus according to claim 1, wherein the feature amount of the binary image is a feature amount of a pixel area calculated by the calculation unit.   The calculating means detects, as the pixel area, a white pixel area surrounded by black pixels or a black pixel area surrounded by white pixels in each of a plurality of binary images created by the creating means. The image processing apparatus according to claim 2, wherein:   The feature amount of the pixel area is at least one of a total number of the pixel areas in a unit of a character, a character string, or a page, rectangular information circumscribing the pixel area, and vector information of the pixel area. The image processing apparatus according to claim 2, further comprising:   The selection unit is configured such that a difference in a feature amount of a pixel region from a binary image created from the input image is smaller than a predetermined threshold among a plurality of binary images created by the creation unit. The image processing apparatus according to any one of claims 2 to 4, wherein a binary image having a minimum resolution is selected.   The said creation means performs a color reduction process with respect to the said input image and the said low resolution image, and produces the binary image for every color from the said color reduction image. The image processing apparatus according to item. An image processing method for generating compressed data from an input image,
Using the input image to generate one or more low-resolution images having a resolution different from the input image;
Creating a binary image for at least two of the input image and the one or more low resolution images;
A selection step of selecting a binary image from the plurality of binary images based on a difference in feature amount between the plurality of binary images created in the creation step;
And a compression step of generating compressed data for the input image by compressing the selected binary image. Computer
Generating means for generating one or a plurality of low resolution images having a resolution different from that of the input image using the input image;
Creating means for creating a binary image for at least two of the input image and the one or more low-resolution images;
Selection means for selecting a binary image from the plurality of binary images based on a difference in feature amount between the plurality of binary images created by the creation means;
A program for functioning as a compression unit that generates compressed data for the input image by compressing the selected binary image.

Images