JP2010056827A - Apparatus and program for processing image - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To optimize a balance between a readability (decrypting performance) and a file size of a document image. <P>SOLUTION: The apparatus for processing an image has: a layout analyzing part 13 for acquiring layout information of the document image stored in an image memory 12; a separation-object-region-determination part 14 for determining a foreground-background separation object region from the document image with the use of the layout information; a character size estimation part 15 for estimation the character size in the region with the character included out of the foreground-background separation object region; a resolution conversion part 16 for varying the resolution of the region in response to the character size; a region separation part 17 for separating the region with the resolution converted and the foreground-background separation object region from the original document image; a region image generation part 18 for generating the foreground image and the background image from the separated region; an image compression part 19 for compressing the foreground image and the background image; and an image integration part 20 for generating the document image arranged in accordance with the layout information. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an image processing apparatus and an image processing program, for example.

  With the proliferation of multifunction peripherals (MFPs) and color scanners and the colorization of documents, the opportunity to store document images in color is increasing. Since a color image tends to have a large file size, the image is often compressed and stored.

  Normally, JPEG compression is often used for color image compression. However, since the document image contains many high-frequency components due to the outline of characters, the compression efficiency of JPEG compression is poor, and image quality improves when the compression ratio is increased. There was a problem of getting worse.

  Therefore, the character component and other components (images from which characters are removed) are separated from the document image as a foreground component and a background component, respectively, and are compressed by different compression methods. A compression technique has been developed (see, for example, Patent Document 1). This technique is referred to as a “compression technique using foreground / background separation”.

  In this compression method using foreground / background separation, character regions are extracted, and for regions determined to be characters, characters are often monochromatic at least in character units. Compress with compression method.

  An image (background component) obtained by interpolating pixels extracted as a character region with surrounding pixel values and removing characters becomes a smooth image, and the compression efficiency of JPEG compression or the like increases. Therefore, the image is compressed by a JPEG compression method or the like.

In that case, it is possible to further increase the compression rate by lowering the resolution of the background component (reducing the image). By displaying the foreground component superimposed on the background component generated in this way, an image that hardly changes in appearance from the original image can be converted into an image having a small file size.
Japanese Patent No. 2611012

  When performing image compression using a compression method based on foreground / background separation, in character areas where the character size is very small or the resolution of the input image is relatively low, the character shape is crushed when the foreground component is reduced, and the characters are collapsed. However, there is a problem that image quality deteriorates such that the readability of the text is impaired or cannot be read. On the other hand, making a large character as high as a small character causes a problem such as overspec or enlargement of the file size.

  The present invention has been made to solve such a problem, and an image processing apparatus and an image processing program capable of optimizing the balance between character readability (readability) of a document image and file size. The purpose is to provide.

  In order to solve the above problems, an image processing apparatus according to the present invention includes an image memory that holds a document image in which characters are written, and a layout analysis performed on the document image held in the image memory. A layout analysis unit that obtains layout information including a position of an element constituting the document image and an attribute indicating the type of the element; and an area to be separated from the document image using the layout information analyzed by the layout analysis unit. A region determining unit to determine, a region including a resolution conversion target element from the attribute among the regions determined by the region determining unit, a size estimation unit that estimates a size of an element of the selected region, A resolution conversion unit that performs resolution conversion of the region in accordance with the size of the element estimated by the size estimation unit; and a region that has undergone resolution conversion by the resolution conversion unit. A separation unit that separates the image from the document image, an image generation unit that generates a foreground image and a background image by performing processing including binarization on the image separated by the separation unit, An image compression unit that individually compresses the foreground image and the background image generated by the image generation unit, and a document image in which the foreground image and the background image compressed by the image compression unit are arranged according to the layout information And an image integration unit.

  The image processing program of the present invention is an image processing program for causing an image processing apparatus to execute processing. The image processing apparatus includes: an image memory that holds a document image in which characters are written; and a document image that is held in the image memory. The layout analysis unit obtains layout information including the position of the element constituting the document image and the attribute indicating the type of the element by performing layout analysis on the layout image, and the layout information analyzed by the layout analysis unit An area determining unit that determines an area to be separated from the document image, and an area that includes an element to be converted from the attribute among the areas determined by the area determining unit; A size estimation unit that estimates the size of the region, and resolution conversion of the region according to the size of the element estimated by the size estimation unit. By performing processing including binarization on the resolution conversion unit, the separation unit that separates the image of the region that has been resolution-converted by the resolution conversion unit from the document image, and the image separated by the separation unit, An image generation unit that generates a foreground image and a background image, an image compression unit that individually compresses the foreground image and the background image generated by the image generation unit, and a foreground compressed by the image compression unit It is made to function as an image integration part which produces | generates the document image which has arrange | positioned the image and the background image according to the said layout information.

  According to the present invention, it is possible to optimize the balance between the readability (readability) of characters in a document image and the file size.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing a configuration of an image processing apparatus according to an embodiment of the present invention.

  As shown in FIG. 1, the image processing apparatus includes an image capturing apparatus 1 such as a digital camera or an image scanner, an image output apparatus 3 such as a printer such as a printing apparatus, a monitor such as a display apparatus, and the like. And a computer 2 (hereinafter referred to as PC2) to which the image capturing device 1 and the image output device 3 are connected.

  The image capturing device 1 reads an image in which characters are written, for example, a document image, and inputs it to the PC 2. That is, the image capturing device 1 functions as an image acquisition unit that captures an image of a surface of a character recognition target form and reads and acquires the image.

  It is assumed that, for example, characters or the like are entered in the form as predetermined information, and the form image obtained by the image capturing device 1 includes a color image such as RGB, a monochrome image, a gray scale image, and the like. There is something to do. Hereinafter, “entry” includes both the meanings when characters and the like are handwritten and printed.

  The image information may be read by a digital camera or an image scanner, or may be acquired by conversion from a document file. Further, the PC 2 may acquire image information stored in another computer through a network such as a LAN.

  The PC 2 has hardware such as a CPU, RAM, ROM and other memory, an auxiliary storage device such as a hard disk device, an input device such as a keyboard and a pointing device such as a mouse, a display device such as a monitor, and an interface board with an image scanner. is doing.

  The hard disk device of the PC 2 stores programs such as an operating system (hereinafter referred to as OS) and image processing software. The PC 2 functions as an image processing apparatus by reading these programs into the memory by the CPU by automatic activation or activation operation by the user. That is, the functions of the PC 2 are realized by the cooperation of a program such as the OS and image processing software read into the memory and the hardware.

  The PC 2 functioning as an image processing apparatus includes an image memory 12, a layout analysis unit 13, a separation target region determination unit 14, a character size estimation unit 15, a resolution conversion unit 16 as a resolution conversion image generation unit, a region separation unit 17, and a region image. A generation unit 18, an image compression unit 19, an image output device 3, and the like are included. The image memory 12 holds (stores) the document image input from the image capturing device 1.

  The layout analysis unit 13 performs layout analysis of the elements constituting the document image read from the image memory 12, thereby determining the position information of each element on the document and the character attribute for determining whether the element is a separation target. Layout information (information such as in which area of the document image the pixels of each element are distributed) including (characters, figures, images, line drawings, etc.).

  Here, the layout analysis is to extract the elements constituting the document image. The unit character area included in the document, the character line area obtained by integrating them, the character string area, the photo area, It is to obtain the chart area and further their attributes.

  For example, in the case of a character string area, the position, width, height, character line included in the character string area, and the like are extracted as attributes. In the case of a character line, the position, width, and height of the line, a character string area including the character line, a unit character area included in the character line, and the like are extracted as attributes.

  In the case of a unit character area, the character position, width, height, character line area including the character, and the like are extracted as attributes. Layout analysis for document images is a common technique in fields such as character recognition.

  In other words, the layout analysis unit 13 performs layout analysis on the document image held in the image memory, so that the position of the elements (characters, line drawings, photographs, etc.) constituting the document image and the type of each element (character Layout information including attributes indicating line drawings or photographs).

  The separation target region determination unit 14 uses the layout information analyzed by the layout analysis unit 13 and includes a target element (a character string serving as a foreground, a line drawing serving as a background, a photograph, or the like) to be separated from a document image. It functions as a determining unit that determines (foreground / background separation target region).

  That is, the separation target area determination unit 14 determines the foreground / background separation target area in the document image using the layout information analyzed by the layout analysis unit 13.

  The character size estimation unit 15 has attribute information of the element (in this case, a character) for resolution conversion in advance, and the layout analysis result of the foreground / background separation target region determined by the separation target region determination unit 14 An area including an element (in this case, a character) for resolution conversion is selected from the attributes included in the layout information, and the size (how many points) of the element (character) in the selected area is estimated. Here, estimating means calculating the size (point) of the character from the height of the area.

  The resolution conversion unit 16 compares the size of the character estimated by the character size estimation unit 15 with a threshold value (allowable range) set in advance. The resolution is converted to a value such that the character size is within the threshold.

  The resolution conversion unit 16 changes the resolution of the foreground / background separation target area in the direction of increasing the resolution when the character size is smaller than a preset threshold value. The resolution conversion unit 16 changes the resolution of the foreground / background separation target region in the direction of decreasing the resolution when the character size is larger than a preset threshold value.

That is, the resolution conversion unit 16 changes the resolution of the foreground / background separation according to the character size estimated by the character size estimation unit 15.
The area separation unit 17 performs a process of separating the image of the foreground / background separation target area whose resolution has been converted according to the character size by the resolution conversion unit 16 and has a size within an allowable range from the document image.

  The region separation unit 17 separates the foreground and the background from the document image by performing processing including binarization on the image of the separated foreground / background separation target region.

  For example, the region separation unit 17 sets the pixel region having the higher density among the binarized image values as the foreground and the other as the background.

  In order to estimate the color of the foreground image (foreground color) or the color of the background image (background color), the pixel value of the region separated as the foreground or background is referred to the original document image, or the average pixel value for each RGB or The mode is assumed. Note that the technique for estimating the color of the image shown here is an example, and other techniques may be used.

  The region image generation unit 18 generates a foreground image and a background image based on the image separated by the region separation unit 17, the estimated foreground color information, and the estimated background color information.

The image compression unit 19 performs a predetermined compression process on the foreground image and the background image generated by the area image generation unit 18.
For example, the foreground image and the background image may be compressed with the same compression technique, or may be compressed with different compression techniques. Here, it is assumed that compression is performed using different compression techniques. The foreground image is compressed by, for example, the MMR compression technique, and the background image is compressed by, for example, the JPEG compression technique.

  The image integration unit 20 combines and integrates each image (foreground image and background image) compressed by the image compression unit 19 according to the layout information, generates a document image having the same arrangement as the original document image, and generates the generated document The image is output to the image output device 3.

The operation of this image processing apparatus will be described below with reference to FIGS.
An image scanner as the image capturing device 1 captures an image from a document such as a document placed on a reading surface, and generates electronic image data including character information (hereinafter referred to as “document image”).

  Then, the document image captured by the image capturing device 1 is input from the image capturing device 1 to the PC 2 (step S101) and held in the image memory 12. An example of a document image input to the PC 2 is shown in FIG.

  As shown in FIG. 3, it is assumed that the document image 40 includes TITLE, a section bar, a large font character, a photograph (picture), a small font character, and the like.

  Subsequently, the layout analysis unit 13 performs layout analysis on the document image 40 in FIG. 3 (step S102).

  In this case, layout analysis is performed, and as shown in FIG. 4, a TITLE area 41, a section bar area 42, a large font character area 43, a photo (picture) area 44, and a small font character area 45, as shown in FIG. 4. Position coordinates (range) are detected. Information on the layout analysis result is referred to as layout information.

  As the position coordinates (range) of the area, at least two points of the x and y coordinates of the upper left top part and the x and y coordinates of the lower right top part of the rectangular range constituting each area are detected. The coordinates of the four corners of the rectangular range may be used.

  Subsequently, the separation target area determination unit 14 uses the layout information to detect, for example, an area in which characters are written, a photograph and a line drawing area, and determines a foreground / background separation target area (step S103).

  Here, based on the layout information of the document image analyzed by the layout analysis unit 13, an area in the row direction (character line area) in which a plurality of characters are arranged horizontally is set as a separation target area.

  In this example, the character line area is the foreground / background separation target area, but the unit character area or the character string area may be the foreground / background separation target area instead of the character line area. Further, a line drawing area, a ruled line area, and the like may be included in the foreground / background separation target area.

  Subsequently, the character size estimation unit 15 selects a region including characters from the foreground / background separation target region determined by the separation target region determination unit 14, and estimates the character size of the region (step S104). ). An example in which character size estimation is performed on the document image of FIG. 3 is shown in FIGS.

  Here, the character size estimation unit 15 detects the regions 41, 43, and 45 (height and width) where the character determined by the separation target region determination unit 14 exists, and the character is determined for each of the regions 41, 43, and 45. Estimate the size (number of points) of. For example, in the case of the region 43, the character size estimation unit 15 detects a plurality of character string regions E1 to E5 from the pixel distribution of the region 43, the continuity of the pixel arrangement, and the like, as shown in FIG. It is determined (estimated) by comparing the number of character points corresponding to the height with information on the number of character points set in advance.

  Then, the character size estimation unit 15 sequentially assigns character line attributes to the regions E1 to E5 of the character strings, and obtains the position (range) and size of the unit character for each character line. For example, from the vertical width (height) of the character string area E1 that is a character line, as shown in FIG. 6, the positions (ranges) and sizes of the unit character areas A1 to A5... That can be arranged in the line. Ask for.

  In this case, the average value or the frequent value of the character units included in the character line may be used as the estimated value, and the height of the character line may be used as the estimated value.

  In addition, the character size may be estimated for each foreground / background separation target area, or the character size may be estimated by grouping nearby foreground / background separation target areas. It is assumed that the character size is not estimated for an area that does not include characters such as a diagram area and a ruled line area.

  Subsequently, the resolution conversion unit 16 converts the resolution of the foreground / background separation target region according to the character size estimated by the character size estimation unit 15 (step S105). In this resolution conversion process, the character size is compared with a preset threshold (range), and the size relationship (within the threshold range or out of range, and when out of range, above or below the range) Therefore, the resolution is not changed, and the resolution is increased or decreased.

  That is, it is estimated that the threshold of the character size for determining whether to perform resolution conversion by enlargement is the threshold Tu, and the threshold of the character size for determining whether to perform resolution conversion by reduction is the threshold Td. If the foreground character size Te is threshold Te <threshold Tu, the resolution is increased, and if the estimated character size Te is threshold Te> threshold Td, the resolution is decreased.

  The threshold values Tu and Td may be fixed values, may be varied according to the application use and the size required to achieve the required character recognition quality, or may be determined dynamically. For example, when the average size of characters included in the document is Ta, the threshold values Tu and Td are dynamically determined by determining values such as Tu = Ta / 2, Td = Ta × 3, and the like. Is possible.

  As the resolution setting method, assuming that the resolution R of the image before conversion and the resolution after conversion are R ′, the value obtained by adding the value Ra to the resolution of the original image is set as the resolution after conversion (R ′ = R + Ra). The original image resolution multiplied by a value Rm is used as the converted resolution (R ′ = R × Rm), and the converted character size is set to a certain value Tt (R ′ = R × Tt). / Te) and the like may be determined by an appropriate method.

  After the resolution is set, enlargement or reduction processing of (R ′ / R) times is performed on the image of the conversion target area. As an enlargement / reduction processing method, a commonly known pixel value interpolation method such as bilinear or bicubic is used.

  Here, if the threshold value Tu is set in advance to 8 points, for example, in the example of FIG. 7, the lower right 6 point character size area EE is subject to resolution conversion by enlargement, and the upper 30 points. The character size area AA is subject to resolution conversion by reduction.

  Subsequently, the region separation unit 17 separates the region whose resolution has been converted according to the character size by the resolution conversion unit 16 or the foreground / background separation target region which has not been subjected to resolution conversion within the threshold range (step S106).

  First, the region separation unit 17 binarizes the image of the foreground / background separation target region. For the binarization of the image, a widely known binarization method such as the binarization method of Otsu is used.

  Subsequently, the region separation unit 17 determines which of the binarized region image values (0 or 1) should be the foreground or the background. Documents are often written with dark contents such as characters and images on a light background such as white. Therefore, the region separation unit 17 separates the pixel region having the higher density among the binarized image values as the foreground and the other as the background.

  In addition, since the area (number of pixels) of the background (the part other than the character) is usually larger than the area of the foreground (the black pixel part of the character) in the area including the character, it is binarized. A pixel area corresponding to a smaller area among the image values may be selected as the foreground.

  In addition, since the edge of the area constituting the character string often becomes the background, the value that occupies the edge of the area constituting the character string area is selected as the foreground among the binarized image values. As a result, foreground / background separation processing that can also handle inverted characters may be performed.

  Here, the region separation unit 17 refers to the original document image for the pixel value of the region separated as the foreground, and estimates the foreground color (foreground color). In order to estimate the foreground color, the pixel value of the region separated as the foreground may be determined by referring to the original document image and taking the average pixel value or mode value for each RGB.

  Subsequently, the region separation unit 17 estimates the background color by referring to the original document image for the pixel value of the region separated as the background. When estimating the background color, the pixel value of the background portion may be an average pixel value or mode value for each RGB with reference to the original image.

  Subsequently, the region separation unit 17 refers to the pixel value corresponding to the background from the original document image and estimates the background color (background color). As a background color estimation method, the original document image stored in the image memory 12 is read out, and an average pixel value or mode value for each RGB is obtained for the background portion, thereby obtaining a background color (background color). ).

  Subsequently, the region image generation unit 18 generates a foreground image and a background image with estimated colors for each of the foreground and background images separated by the region separation unit 17 (step S107).

  The foreground image is generated by integrating the foreground pixel components of the foreground / background separation target region having the same resolution of the foreground and the same resolution. When integrating the foreground image and the background image, arrangement information (placement information indicating which foreground image is integrated into which position of the background image) obtained by layout analysis in advance is used.

  Thereby, when the foreground image and the background image are integrated, the foreground image can be arranged at a correct position. A background image can be generated by replacing the pixel value of a pixel constituting the foreground from the original document image with the background color of the foreground including the pixel.

  By this processing, the background image becomes a smooth image that does not include characters and the like, and the image efficiency can be improved. When it is desired to reduce the size of the background image, resolution conversion for reducing the background image may be performed. An example of the generated foreground / background image is shown in FIG.

  Subsequently, the image compression unit 19 performs compression processing on each of the foreground image and the background image generated by the region image generation unit 18 (step S108), and as shown in FIG. 9, the foreground images having different resolutions are obtained. 91, 92, 93 and a background image 94 are generated. Note that when the character size is within the threshold value, resolution conversion may not be performed, and the respective resolutions are not necessarily different.

  The foreground image 91 is generated with a resolution of 200 dpi, RGB (0, 0, 100), coordinate information x1, and the like. The foreground image 92 is generated with a resolution of 300 dpi, RGB (0, 0, 0), coordinate information x2, and the like. The foreground image 93 is generated with a resolution of 600 dpi, RGB (0, 0, 0), coordinate information x3, and the like. The background image 94 is generated with a resolution of 150 dpi or the like.

  The foreground images 91, 92, 93 are compressed by MMR compression suitable for binary image compression, and the background image 94 is compressed by JPEG compression suitable for natural image compression. Thereby, a compression rate can be raised. The compression method used here is not limited to MMR compression and JPEG compression, and other compression methods may be used.

  Subsequently, the image integration unit 20 integrates (rearranges) the foreground images 91, 92, 93 and the background image 94 compressed by the image compression unit 19, as shown in FIG. Is generated and output to the image output device 3 (step S109).

  Here, a case where Portable Document Format (hereinafter referred to as “PDF”) is selected as the output format will be described. PDF is a format in which a plurality of image formats and resolutions can be mixed, and a mask attribute can be assigned to a binary image.

  That is, the image integration unit 20 generates a document image in which a foreground image is arranged on a background image, and gives a mask attribute for each pixel to the foreground image. By displaying the foreground color set in each foreground image only for pixels whose mask is ON (1), it is possible to output a document with the same appearance as the original image from the foreground image and the background image.

  As described above, according to the image processing apparatus of this embodiment, when a small character is included in a document image, it is possible to generate an easy-to-read document with little deterioration in image quality despite a small file size. it can.

  Here, the image quality improvement effect according to the present embodiment will be described with reference to FIGS. FIG. 11 shows an image when the original color image is binarized as it is, and FIG. 12 shows an image when the resolution is improved by this embodiment.

  As shown in FIG. 11, when a 300 dpi color image 95 (hereinafter referred to as “original image 95”) in the foreground portion is binarized at the same resolution, gradation information due to binarization is lost, and thus binarization is performed. In the small characters of the image 96, the visual image quality is greatly deteriorated. The outline shape of characters and the thickness of character strokes also appear to be irregular.

  On the other hand, as shown in FIG. 12, in the image processing apparatus of this embodiment, resolution conversion (300 dpi → 600 dpi) is performed in the direction of increasing the resolution from the original image 95 to generate a high resolution image 97. In the case of binarization, the binarized image 98 has little deterioration in the appearance image even if gradation information is missing.

  Comparing each other's images (the image 96 in FIG. 11 and the image 98 in FIG. 12), the image 98 generated by the image processing apparatus of the present embodiment (the foreground image is separated into the foreground and the background, the resolution is converted, and then integrated) The effect of improving the image quality is obvious.

  In the image processing apparatus of the present embodiment, the foreground component file size may be increased by improving the resolution of the foreground including characters, but the foreground component compressed document image is compressed using the MMR compression technique. In this case, the increase in the file size is small compared with the increase in the number of pixels, and it can be said that the merit of improving the image quality is larger.

  In addition, when large characters are included in the image, even if the resolution of the foreground component in that area is lowered, there is little effect on the image quality.Therefore, by performing resolution conversion in the direction of decreasing the resolution, the output image The file size can be made smaller.

  In addition, this invention is not limited only to the said embodiment, You may deform | transform a component in the range which does not deviate from the summary in an implementation stage. Moreover, various inventions can be configured by appropriately combining a plurality of components disclosed in the embodiment.

  For example, some components may be deleted from all the components shown in the embodiment. Furthermore, constituent elements over different embodiments may be appropriately combined.

1 is a block diagram illustrating a configuration of an image processing apparatus according to an embodiment of the present invention. It is a flowchart which shows operation | movement of an image processing apparatus. It is a figure which shows an example of the document image of a process target. It is a figure which shows each area | region detected by the layout analysis. FIG. 4 is a diagram illustrating an example in which a character string region is detected with respect to a character region of the document image in FIG. 3. It is a figure which shows the example which detected the unit character area | region contained in the character string area | region of FIG. It is a figure which shows the designated character size of each area | region of a document image. It is a figure which shows a foreground image. It is a figure which shows the foreground image and background image which converted the resolution with respect to the original document image. It is a figure which shows the document image produced | generated by the substantially same arrangement | positioning as the original sentence image. It is a figure which shows the image quality when binarizing the original image of a color as it is. It is a figure which shows the image quality in the case of binarizing after performing resolution conversion with respect to the color original image.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Image capture device, 2 ... Computer (PC), 3 ... Image output device, 12 ... Image memory, 13 ... Layout analysis part, 14 ... Separation object area | region determination part, 15 ... Character size estimation part, 16 ... Resolution conversion , 17... Region separation unit, 18... Region image generation unit, 19... Image compression unit, 20.

Claims (6)

An image memory for holding a document image filled with characters;
A layout analysis unit that obtains layout information including a position of an element constituting the document image and an attribute indicating the type of the element by performing a layout analysis on the document image held in the image memory;
An area determination unit that determines an area to be separated from the document image using layout information analyzed by the layout analysis unit;
Among the areas determined by the area determining unit, a region including a resolution conversion target element is selected from the attribute, and a size estimating unit that estimates the size of the element in the selected region;
A resolution conversion unit that performs resolution conversion of the region according to the size of the element estimated by the size estimation unit;
A separation unit that separates an image of a region whose resolution has been converted by the resolution conversion unit from the document image;
An image generation unit that generates a foreground image and a background image by performing processing including binarization on the image separated by the separation unit;
An image compression unit that individually compresses the foreground image and the background image generated by the image generation unit;
An image processing apparatus comprising: an image integration unit that generates a document image in which a foreground image and a background image compressed by the image compression unit are arranged according to the layout information. The resolution converter
The image processing apparatus according to claim 1, wherein when the size of the element estimated by the size estimation unit is smaller than a preset threshold, the resolution of the region is changed in a direction of increasing the resolution. The resolution converter
The image processing apparatus according to claim 1, wherein when the size of the element estimated by the size estimation unit is larger than a preset threshold, the resolution of the region is changed in a direction of decreasing the resolution. In an image processing program for causing an image processing apparatus to execute processing,
The image processing apparatus;
An image memory for holding a document image filled with characters;
A layout analysis unit that obtains layout information including a position of an element constituting the document image and an attribute indicating the type of the element by performing a layout analysis on the document image held in the image memory;
An area determination unit that determines an area to be separated from the document image using layout information analyzed by the layout analysis unit;
Among the areas determined by the area determining unit, a region including a resolution conversion target element is selected from the attribute, and a size estimating unit that estimates the size of the element in the selected region;
A resolution conversion unit that performs resolution conversion of the region according to the size of the element estimated by the size estimation unit;
A separation unit that separates an image of a region whose resolution has been converted by the resolution conversion unit from the document image;
An image generation unit that generates a foreground image and a background image by performing processing including binarization on the image separated by the separation unit;
An image compression unit that individually compresses the foreground image and the background image generated by the image generation unit;
An image processing program that functions as an image integration unit that generates a document image in which a foreground image and a background image compressed by the image compression unit are arranged according to the layout information. The resolution converter
The image processing program according to claim 4, wherein when the size of the element estimated by the size estimation unit is smaller than a preset threshold, the resolution of the region is changed in a direction of increasing the resolution. The resolution converter
5. The image processing program according to claim 4, wherein when the size of the element estimated by the size estimation unit is larger than a preset threshold value, the resolution of the area is changed in a direction of decreasing the resolution.

Images