JP2004032554A - Image processor and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processor capable of separating an image wherein a character / line drawing and a pattern photographing image are intermingled into the character / line image part and the pattern part with high separation performance at high speed according to the local property of the image. <P>SOLUTION: A first attribute separation section 4 performs attribute separation by comparatively simple processing such as the discrimination of a pixel close to a black color to be a character / line image part. A second attribute separation section 5 uses comparatively sophisticated processing such as statistical processing to distinguish the character / line image part from other parts. A user uses an area designation section 2 to designate the area of an object of the sophisticated separate processing by the second attribute separation section 5 among image data received from an image data input section 1. A changeover section 3 shares and supplies the data in the area designated by the area designation section 2 to the second attribute separation section 5 and data of the other areas to the first attribute separation section 4 respectively among the received image data. Since the section 5 can separate the data which the first attribute separation section 4 can hardly separate, the image processor can obtain an excellent attribute separation result. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to an image processing apparatus and method for processing image data in which characters / line drawings and pictures are mixed.
[0002]
[Prior art]
In recent years, ITU-T Recommendation T. In an MRC (Mixed Raster Content) defined in 44, image data in which a character / line drawing and a continuous tone pattern such as a photograph are mixed is indicated by a background plane, a foreground plane, or both of them. The system is divided into a total of three layers, that is, a selection plane, and a method for individually compressing and encoding each plane is defined. This ITU-T Recommendation T. According to 44, the character / line drawing portion and the picture portion can be separated into separate planes, and each plane can be compression-encoded by a compression encoding method suitable for the characteristics of the image data included in each plane. For example, the plane of the picture portion is in accordance with ITU-T Recommendation T. 81, compression encoding is performed by a lossy compression method such as the JPEG method, and the plane of the character / line drawing portion is in accordance with ITU-T Recommendation T.81. For example, compression encoding is performed by a lossless compression method such as the MMR method shown in FIG.
[0003]
As described above, the technique of separating image data in which characters / line drawings and pictures are mixed into three layers of image data planes according to the local characteristics and selectively switching the compression encoding method for each plane is disclosed in JPEG. There is an advantage that the compression image quality can be kept high while the compression efficiency is improved as compared with the case where the text / photo mixed image data is compressed by the method alone.
[0004]
On the other hand, how to separate a character / line drawing portion from a picture portion is described in ITU-T Recommendation T. 44, not standardized. Therefore, as an attribute separation method for generating the MRC format, a simple method of performing separation on a pixel basis based on a pixel value is disclosed in, for example, JP-A-2001-223903 and JP-A-2001-127999 by the present applicant. Various methods have been proposed, up to the complicated one.
[0005]
[Problems to be solved by the invention]
However, the attribute separation technology disclosed as a conventional technology has the following problems.
[0006]
First, a relatively simple attribute separation algorithm can quickly separate an image in which characters and line drawings and pictures are mixed, but the separation capability is not so good. For example, characters and lines that are superimposed on a pattern part It is difficult to separate the parts well.
[0007]
On the other hand, with a complicated separation algorithm, it is generally possible to separate an image having a complicated structure since the separation performance is high.For example, it is possible to satisfactorily separate a character / line drawing portion superimposed on a picture portion. On the other hand, the processing becomes complicated and the processing speed decreases.
[0008]
The present invention solves the above-mentioned problems of the prior art. In the present invention, an image in which a character / line drawing and a pattern photographic image are mixed is converted into a character / line drawing portion and a pattern with high speed and high separation performance according to the local property. It is an object of the present invention to provide an image processing device that can be separated into parts.
[0009]
[Means for Solving the Problems]
In order to achieve the above object, an apparatus according to the present invention accepts a plurality of attribute separating means for separating image data into a character / line drawing part and a picture part by different methods, and receives designation of area division for an image to be processed. And control means for causing the corresponding attribute separating means to process data of each area of the image to be processed in accordance with the area division designation.
[0010]
Further, the apparatus according to the present invention selects the image data from first foreground image data, first background image data, and any one of the first foreground image data and the first background image data. A first selection data indicating whether each pixel is to be divided into three layers: first selection data indicating whether the image data is divided into three layers, and second foreground image data in a different method from the first separation means. A second background image data, and second selection data indicating which of the second foreground image data and the second background image data is to be selected. Receiving the designation of area division for the image to be processed, and distributing the data of each area of the image to be processed to the first separation means and the second separation means according to the area division designation. Control means for performing the processing.
[0011]
According to a preferred aspect of the present invention, the first selection data generated by the first separation unit and the second selection data generated by the second separation unit are combined to generate combined selection data. Means.
[0012]
In another preferred aspect of the present invention, the third selection data is generated based on the area specifying information output by the area specifying means, and the three-layer data output by the first separating means includes: The image processing apparatus further includes a multi-layer image data generation unit that generates a multi-layer image data plane including the three-layer data output by the second separation unit and the third selection data.
[0013]
In addition, the apparatus according to the present invention includes a first attribute separating unit that separates image data into a character / line drawing portion and a picture portion, and a display that displays information related to a separation result output by the first attribute separating unit. Means, area designation means for receiving designation of a closed area on the image data in association with the display of the display means, and image data included in the closed area designated by the area designation means, There is provided a second attribute separating unit that separates a character / line drawing part and a picture part by a method different from the attribute separating unit.
[0014]
BEST MODE FOR CARRYING OUT THE INVENTION
FIG. 1 is a diagram showing a schematic configuration of an image processing apparatus according to the present invention.
[0015]
The image data input unit 1 is means for taking in image data in which a character / line drawing and a pattern are mixed into the image processing apparatus. As an example of the image data input unit 1, for example, a scanner that optically reads an original placed on a platen glass can be cited.
[0016]
The area specifying unit 2 is a means for specifying a closed area at an arbitrary position on the image data taken in from the image data input unit 1. As the area specifying unit 2, for example, a configuration in which the position and size of the area are input in a numerical format from a keyboard can be used. Further, a configuration in which an area is visually and intuitively specified using a pointing device such as a touch pen or a mouse on an image displayed on the display may be used.
[0017]
The switching unit 3 receives area information indicating the area specified by the user from the area specifying unit 2 and converts the image data captured by the image data input unit 1 into a first attribute separating unit 4 or a second To the attribute separation unit 5 of the application.
[0018]
The first attribute separating section 4 and the second attribute separating section 5 are means for separating input image data in which a character / line drawing and a pattern are mixed into image data belonging to a character / line drawing and image data belonging to a pattern. It is. The first attribute separating unit 4 and the second attribute separating unit 5 use different attribute separating methods. Here, as the first attribute separating unit 4, a unit which does not have a very high separating performance but can separate a character / line drawing portion and a picture portion at a high speed by relatively simple processing is used, and the second attribute separating unit 4 is used. As the separation unit 5, a separation unit having a low processing speed but high separation performance is used. The first attribute separating unit 4 can use a configuration in which, for example, a local property of image data is analyzed to separate a character / line drawing portion from a picture portion. The second attribute separating unit 5 may employ a configuration in which, for example, the entire image data is regarded as a document and the structure is analyzed to separate the character / line drawing portion from the picture portion.
[0019]
An outline of the entire operation will be described with reference to FIGS. FIG. 3 shows an example of image data in which a character / line drawing part and a picture part read from the image data input unit 1 are mixed. This image includes a portion 31 where a pattern and a character overlap, a portion 32 consisting only of characters and line drawings, and a portion 33 consisting only of a pattern.
[0020]
In general business documents, the color of characters and lines is often black, and by paying attention to this point, many characters and line drawings can be extracted by detecting "black" pixels of image data. . Therefore, in the following example, the first attribute separating unit 4 is a unit using such a simple attribute separating algorithm.
[0021]
On the other hand, even in a business document, the color of characters or lines to be emphasized may be other than black, or a picture may be arranged in the background of a character / line drawing portion. Such a portion is difficult to be separated by the first attribute separating unit 4. Such a portion can be separated into a character / line drawing portion and a picture portion by using, for example, an algorithm for analyzing local properties of image data by statistical processing or performing document structure analysis. . Here, the second attribute separating unit 5 uses such a complicated attribute separating algorithm.
[0022]
The area specifying unit 2 is used to specify a part that requires complicated processing (that is, a part to be processed by the second attribute separating unit 5) in image data in which characters / line drawings and pictures are mixed. When processing the image illustrated in FIG. 3, the user uses the area specifying unit 2 to specify a portion 33 where the picture and the character overlap each other as a closed area.
[0023]
The switching unit 3 sends the image data in the area corresponding to the part 33 where the picture and the character overlap in the image data to be processed to the second attribute separating unit 5 based on the area information output from the area specifying unit 2. , And the image data of the other areas are distributed to the first attribute separating unit 4.
[0024]
The first attribute separating unit 4 and the second attribute separating unit 5 respectively convert the image data input from the switching unit 3 into a character / line drawing portion and a picture portion according to a predetermined attribute separating algorithm of the image data. To separate.
[0025]
In the above example, an example in which the image data is distributed to the two attribute separating units 4 and 5 by switching by the switching unit 3 has been described, but such a switching unit 3 is not necessarily required. For example, a configuration is adopted in which data of an input image input from the input unit 1 is input in parallel to both the first attribute separating unit 4 and the second attribute separating unit 5, and the area information output by the area specifying unit 2 Accordingly, a configuration in which one of the separation units 4 and 5 is selectively operated is also possible. For example, the output of the area specifying unit 2 is supplied to both of the separating units 4 and 5, and the area specifying unit 2 synchronizes with the supply of the image data from the input unit 1 to the separating units 4 and 5. The region information (ie, a signal indicating whether or not the pixel belongs to the designated closed region) according to the supplied pixel data is supplied to both of the separation units 4 and 5. May be controlled to operate or stop according to the area information.
[0026]
As described above, since a plurality of attribute separation algorithms can be selectively switched based on the area information output from the area specifying unit 2, most of the image is processed by the attribute separation algorithm capable of high-speed processing. Only in the area where performance is required, processing can be performed by an attribute separation algorithm having a low processing speed but high separation performance. In other words, the attribute of the entire image can be accurately separated without significantly lowering the processing speed.
[0027]
Next, the configurations of the first attribute separating unit 4 and the second attribute separating unit 5 will be described with reference to FIG. The first attribute separating unit 4 and the second attribute separating unit 5 both have the module configuration shown in FIG. 2, but the processing contents in the character shape extracting unit 21 are different between them. In the following, a case will be described as an example where attribute separation is performed by focusing on the shape of a character / line drawing portion. However, this is merely an example, and other methods may be used, such as extracting a character area with a set of circumscribed rectangles of character parts.
[0028]
The character shape extraction unit 21 extracts a shape such as a character or a line drawing from the input image data (an extraction method will be described later), and assigns “1” to the extracted character / line drawing part and “1” for the character / line drawing part. "0" is output for a portion corresponding to the background (ie, outside the extracted shape). The output of the character shape extracting unit 21 is a selection for uniquely selecting one of the first image data DT1 (corresponding to the character color data in the following description) and the second image data DT2 (corresponding to the background image data in the following description). It can be used as data SEL. Note that the color space of the image data to be input to the character shape extraction unit 21 is CIE-L in the following description. ^* a ^* b ^* Although a uniform color space is assumed, the present invention is not limited to this.
[0029]
The multiplexer (MUX) 22 has two systems of outputs: character color data DT1 and background image data DT2. The multiplexer 22 has a processing target L ^* a ^* b ^* Image data and selection data SEL from the character shape extraction unit 21 are input. Then, when the selection data indicates “1”, the multiplexer 22 outputs the current L (that is, of the pixel corresponding to the selection data). ^* a ^* b ^* At the same time as outputting the value of the image data as the character color data DT1, as the background image data DT2, image data of a predetermined value, for example, L ^* = A ^* = B ^* = 0 is output. When the selection data indicates “0”, the multiplexer 22 outputs the input L at that time. ^* a ^* b ^* At the same time as outputting the value of the image data as the background image data DT2, image data of a predetermined value, for example, L ^* = A ^* = B ^* = 0 is output. Therefore, the selection data SEL, the character color data DT1, and the background image data DT2 output from the first attribute separating unit 4 and the second attribute separating unit 5, respectively, have the same size in the main scanning direction and the sub-scanning direction. These three outputs are referred to in ITU-T Recommendation T. 44 (MRC) can be used as three planes.
[0030]
The result of applying the above-described attribute separation by character shape extraction to image data in which a character portion and a picture portion are mixed will be described with reference to FIG.
[0031]
FIG. 6A shows an example of an input document image 60 in which "Japan map" corresponding to a picture portion and "JAPAN" corresponding to a colored character portion are mixed. On the other hand, FIG. 6B is a diagram showing three image planes 61, 62, and 63 formed as a result of the attribute separation processing on the original image. The character shape of “JAPAN” extracted by the attribute separation processing becomes the plane 61 (selected plane) of the selected data SEL. Also, in the plane 62 (character color plane) of the first image data DT1, the selection data SEL of the document image 60 corresponds to “1” (ie, the “black” portion of the selection data SEL in FIG. 6B). The data of the area to be selected is selected. In this case, the data is a colored character "JAPAN". Then, in the plane 63 (background image plane) of the second image data DT2, data of an area where the selection data SEL is “0” in the document image 60 is selected. Therefore, in the example of FIG. 6, the figure of the plane 63 “Japan map” of the second image data DT2 is selected.
[0032]
The three planes shown in FIG. 6B obtained as a result of the attribute separation may be directly used as the three image data planes of the MRC. However, considering the data compression efficiency, there is room for improvement in the character color plane and the background image plane. There is. FIG. 6C shows a set of image data planes having better compression efficiency. In this example, the same selection plane 66 as that of the example in FIG. 6B is used, but the image pattern of the character color plane 64 and the background image plane 67 is simplified.
[0033]
That is, since the shape of the character / line drawing portion is represented on the selection plane 66, the character color plane only needs to have 64, the color information of the character / line drawing portion. Absent. For this reason, if instead of the character color plane 61 shown in FIG. 6B, a character plane 64 in which an area containing each character is represented by a patch 65 of the character color is used, the original document image 64 The character part can be reproduced. From the viewpoint of data compression efficiency, it is preferable to use a patch 65 having a simple shape. In the example of FIG. 6C, a rectangular area in which each side is parallel to the main scanning direction and the sub-scanning direction and circumscribes each character is used as the character color patch 65. By using a plane 64 composed of patches 65 of a simple shape as a character color plane, the edge information is significantly less than that of the plane 61 including the character shape itself. Data compression can be performed efficiently by the JPEG method shown in FIG.
[0034]
The following improvements can be made to the plane 67 of the background image. That is, in the output of the background image DT2 of the multiplexer 22, the value in the area of the character / line drawing portion of the document image 60 is 0, and no significant image data remains. A flattening process is performed based on the image data. That is, each pixel in the area is replaced with the value of a pixel around the area. This makes it possible to perform data compression with high efficiency because the outer edge of the character, which causes a reduction in the compression ratio in the JPEG system, becomes gentler. In this way, a multilayer image data plane as shown in FIG. 6C is generated.
[0035]
Next, the processing contents of the character shape extraction unit 21 for each of the first attribute separation unit 4 and the second attribute separation unit 5 will be described.
[0036]
FIG. 4 is a flowchart illustrating a simple attribute separation algorithm executed by the character shape extraction unit 21 of the first attribute separation unit 4 to extract a “black” portion of image data as a character / line drawing portion. In this process, the image data is sequentially processed in a predetermined order such as a raster scanning order. Each time data of one pixel is input according to this order, first, the luminance component L ^* Is compared with a predetermined threshold value τ1 (S41). Here, since a low-luminance pixel close to black is detected as a character / line image portion, a small predetermined value is set as the threshold τ1 in advance. If the luminance signal L of the pixel ^* Is larger than τ1, it is determined that the pixel is not black, that is, it is not part of a character / line drawing, and the selection data SEL for the pixel is set to “0” (S42). On the other hand, the luminance signal L of the pixel ^* Is less than or equal to τ1, the color difference component a ^* , B ^* Is determined based on the size of the image. Since the color difference component can take a negative value, the sum of the squares is calculated here and compared with a predetermined threshold value τ2 (S43). Since a color close to black has a small color difference component, a relatively small predetermined value is set as the threshold τ2 in advance. Color difference component a ^* , B ^* If the sum of the squares is smaller than τ2, it is determined that the color is a color close to black, that is, a part of a character or a line drawing, and the selection data for the pixel is set to “1” (S44). When the sum of squares of the color difference components is equal to or greater than τ2, it is determined that the pixel is not black, that is, it is not a part of a character or a line drawing, and the selection data for the pixel is set to “0” (S42). This process is repeated for all the image data (S45). Through the above processing, it is possible to obtain the selection data SEL in which a portion close to black in the image data indicates a binary “1”.
[0037]
As is clear from the above description, the present algorithm executed by the first attribute separation unit 4 is a simple process, so that it is easy to implement hardware by ASIC or the like, and even if it is implemented by software, Processing is possible.
[0038]
Next, an example of an algorithm executed by the second attribute separating unit 5 that is more complicated than that of the first attribute separating unit 4 will be described with reference to FIG.
[0039]
First, the L to be processed ^* a ^* b ^* L of image data ^* , A ^* , B ^* For each component, the image data of the component is divided into blocks of a predetermined size (for example, 8 × 8 pixels) (S51), and a histogram of the pixel values of the component of each pixel in the block is obtained. Then, a difference between the maximum pixel value Max and the minimum pixel value Min is obtained, it is checked whether the difference is larger than a predetermined first threshold value TH1, and a variance σ of the pixel values in the block is obtained. It is checked whether the variance value σ is greater than a predetermined second threshold value TH2 (S52). If it is determined that both the difference between the pixel values and the variance value are larger than the corresponding threshold values TH1 and TH2, it means that the block includes a portion where the pixel value change is steep, and thus the character / line drawing portion is included. It is determined to be a block (referred to as a character / line drawing block), and the process proceeds to step S53. In step S52, if either one of the difference value or the variance value is equal to or smaller than the threshold value, it is determined that the block is not a character / line drawing block, and all the selection data for each pixel in the block is “0”. (S57). It should be noted that the value of the selection data obtained here is not the final result, but is combined with the value of the selection data obtained later in the processing for other components to obtain the final selection data for each pixel.
[0040]
For the block determined to be a character / line drawing block in step S52, an average value of the maximum pixel value Max and the minimum pixel value Min of the block is obtained, and this is set as a threshold value B relating to the binarization processing of the block (S53). ). Then, the pixel value Dij of the pixel (i, j) (i, j is a number indicating the ordinate and abscissa of the pixel in the block) belonging to the block is compared with the threshold value B (S54). “1” is assigned to the pixel having a large value as the selection data SEL (S56), and “0” is assigned to the pixel having a value smaller than the threshold B as the selection data SEL (S55). This process is repeated for all pixels in the block (LOOP1). Thereby, selection data of each pixel in the block for the component is obtained. The selection data obtained in steps S55 and S56 are also temporary values, as in the case of step S57, and are combined with the selection data obtained for another component later.
[0041]
For the same block, the processing of steps S51 to S56 ^* , A ^* , B ^* (LOOP2). Thereby, the value (“0” or “1”) of the selection data of each pixel in the block is obtained for each of these components. For each pixel in the block, the logical sum of the values of the selection data of these three components is obtained, and this is output as the final selection data of the pixel (S58).
[0042]
The above is the processing for one block. This processing for one block is repeated until all the image data for one page is processed (LOOP3). This completes attribute separation for the entire page.
[0043]
In the above description, the luminance signal L of the image data ^* , Color difference signal a ^* , B ^* Was used to determine the final attribute separation result, but this is not essential. For example, the luminance signal L ^* Even in a configuration in which the final attribute separation result is obtained based only on the statistical analysis result, a favorable result can be obtained.
[0044]
As a process of the character shape extracting unit 21, for example, a process of obtaining a spatial frequency distribution of image data by an orthogonal transformation process or the like and extracting a shape of a character / line drawing portion based on the result is used. You can also.
[0045]
An example of the attribute separation algorithm performed by the first attribute separation unit 4 and the second attribute separation unit 5 illustrated in FIG. 1 has been described above. In the configuration shown in FIG. 1, the first attribute separating unit 4 and the second attribute separating unit 5 respectively assign the regions assigned to the attribute separating unit (4 or 5) in accordance with the region specification by the region specifying unit 2. Are output as attribute separation results (ie, three data of selection data, character color data, and background image data). In this configuration, each of the attribute separation units 4 and 5 outputs three output data as the attribute separation result. However, the input image data is divided by the switching unit 3 into the same pixel data. Is input to only one of the attribute separation units 4 and 5, so that the outputs of these two attribute separation units 4 and 5 can be integrated in chronological order, whereby the three planes of the attribute separation result for the entire input image can be integrated. Data can be obtained.
[0046]
According to the above-described embodiment, by performing attribute separation at high speed with a relatively simple process in a region where attribute separation is easy in an image, and by applying an advanced separation process to a region where attribute separation is difficult. For example, advanced attribute separation such as separating characters arranged on a natural image can be performed. According to the present embodiment, when the attribute separation processing is realized by software, advanced attribute separation can be realized without greatly lowering the overall processing speed.
[0047]
Next, another configuration of the device according to the present invention will be described with reference to FIG.
[0048]
In this configuration, the image data input unit 71 is a block that takes in image data in which characters / line drawings and patterns are mixed into the image processing apparatus, and corresponds to the input unit 1 having the configuration in FIG. The area specifying section 72 is a block for specifying a closed area at an arbitrary position on the image data captured by the image data input section 71, and corresponds to the area specifying section 2 having the configuration in FIG. The operator designates a closed area from the area designating unit 72 so as to surround a portion where the character / line drawing and the picture are complicatedly overlapped while referring to a display (not shown). The area information signal output by the area specifying section 72 to the first and second character shape extracting sections 73 and 74 at the subsequent stage is “1” when indicating the inside of the closed area, and “1” when indicating the outside of the closed area. 0 ".
[0049]
The first character shape extraction unit 73 is a block that can execute the attribute separation algorithm described with reference to FIG. 4, for example, which has a low separation performance but a high processing speed. The second character shape extraction unit 74 is a block that can execute the attribute separation algorithm described with reference to FIG. 5, for example, which is not so fast but has high separation performance. The operation of each of the character shape extracting units 73 and 74 is controlled by the logical level of the area information signal output from the area specifying unit 72, and the position where the area information signal indicates “0”, that is, the closed state specified by the operator. Outside the region, only the first character shape extraction unit 73 having a high processing speed can operate. At the position where the area information signal indicates “1”, that is, inside the closed area, only the second character shape extraction unit 74 having high separation performance can operate.
[0050]
As is clear from the description of FIG. 4 and FIG. 5, the output of the first and second character shape extraction units indicates “1” for the pixel classified as character / line drawing and does not belong to the character / line drawing. Since “0” is indicated for a pixel classified as a background such as a picture, one selection data for the entire image is obtained by performing a logical sum on the selection data output by each character shape extraction unit. be able to. The OR circuit 75 performs a logical OR operation on the selection data SEL1 output by the first character shape extraction unit 73 and the selection data SEL2 output by the second character shape extraction unit 74 to obtain final selection data. It is a block.
[0051]
The multiplexer 76 has an input L ^* a ^* b ^* When the selection data indicates "1", the image data is output as character color data, and at the same time, image data having a predetermined value, such as L ^* = A ^* = B ^* = 0 is output. When the selection data indicates “0”, the multiplexer 76 outputs the input image data as background image data, and at the same time, outputs image data of a predetermined value as character color data, for example, L ^* = A ^* = B ^* = 0 is output. ITU-T Recommendation T. According to 44, the selection data described above corresponds to the mask layer, the character color data corresponds to the foreground layer, and the background image data corresponds to the background layer.
[0052]
As described above, even with the configuration of FIG. 7, the same attribute separation result as that of the above-described embodiment described with reference to FIG. 1 can be obtained. In the configuration of FIG. 7, only one multiplexer unit 76 is required.
[0053]
Next, still another configuration of the device according to the present invention will be described with reference to FIGS.
[0054]
ITU-T Recommendation T. 44, in Annex A, there is a description of MRC extended to three or more layers. According to this, in the MRC of three or more layers, layer 2, layer 3,..., Layer N are superimposed in ascending order with the lowest layer as layer 1. In this multilayer structure, the odd-numbered layers are referred to as image layers composed of multi-valued image data, and the even-numbered layers are referred to as mask layers composed of binary selection data. As shown in FIG. 8, the image layer and the mask layer are superimposed on each other with the image layer arranged at the bottom as a background and the mask layer and the image layer superimposed thereon in pairs. It is supposed to be. According to this rule, the result of attribute separation by two different attribute separation algorithms described in the above embodiment can be handled as an MRC of three or more layers. FIG. 9 shows an example of this device configuration.
[0055]
9, an image data input unit 91 is a block that takes in image data in which a character / line drawing and a pattern are mixed into the inside of the image processing apparatus, and corresponds to the input unit 1 in FIG. The area specifying unit 92 is a block that specifies a closed area at an arbitrary position on the image data captured by the image data input unit 91, and corresponds to the area specifying unit 2 in FIG. In the same manner as described above, the operator designates a closed area from the area designating section 92 so as to surround a portion where the character / line drawing and the picture are complicatedly overlapped while referring to a display (not shown) or the like. The area information signal output by the area specifying unit 92 is “1” when indicating the inside of the closed area, and “0” when indicating the outside of the closed area.
[0056]
The first attribute separation unit 93 is a block capable of executing an attribute separation algorithm having a not so high separation performance as described in FIG. 4 but a high processing speed. However, unlike the above-described embodiment, the first attribute separation unit 93 in this example performs attribute separation processing on all input image data regardless of the area information signal of the area specification unit 92.
[0057]
The second attribute separating unit 94 is a block capable of executing an attribute separating algorithm having a high processing performance, although the processing speed is not so high as described in FIG. 5, and the area information signal indicates the inside of the closed area. Attribute separation processing is performed only when the value is “1”.
[0058]
Here, since the attribute separation result output by the second attribute separation unit 94 is valid when the area information signal is “1”, the area information signal is used as the second mask layer (mask 2). Accordingly, when the logic of the second mask layer is “1”, the image data included in the layer above the second mask layer becomes valid, and the result of attribute separation by the second attribute separation unit 94 is displayed thereon. Can be superimposed.
[0059]
In this configuration, the attribute separation results of the image data included in the closed area specified by the area specifying unit 92 are output from the first and second attribute separating units 93 and 94. The output of the first attribute separation unit 93 will include useless image data in the resulting overwritten portion. An optimal method for the attribute separation result for each attribute, specifically, a first background layer (background 1. background image data) or a first foreground layer (foreground 1. foreground image data or character color data) is often used. Value image, the ITU-T Recommendation T. 81, and the first mask layer (selection data) is a binary image. The MH / MR system shown in FIG. 6, the MMR method shown in FIG. In the case of performing compression by the JBIG method or the like shown in FIG. 85, it may not be desirable because remaining useless data acts in a direction in which the compression ratio cannot be increased. In such a case, if necessary, of the attribute separation results of the first attribute separation unit 93, the separation results of the part included in the area specified by the area specification unit 92 are deleted and a predetermined initial value is set. May be added.
[0060]
Next, another embodiment of the present invention using a personal computer will be described with reference to FIG.
[0061]
The scanner 101 as a means for capturing image data in which characters, line drawings and patterns are mixed irradiates a document placed on platen glass with a fluorescent lamp or the like, and optically reflects reflected light using an imaging device such as a CCD. An image reading unit 102 for converting analog image data into digital image data after performing reading, offset adjustment of a read signal, and gain adjustment is internally provided. A signal output from the image reading unit 102 is input to an attribute separating unit 103 inside the scanner 101. The attribute separating unit 103 separates the input image data into a character color data plane, a background image data plane, and a selected data plane as in the above-described embodiment, and outputs a multi-layer image data plane including the three planes from the scanner 101 to the outside. It has a function to output to Here, the attribute separation process executed by the attribute separation unit 103 is desirably a high-speed processing method capable of performing processing in almost real time with respect to the document scanning time of the scanner. In the following description, it is assumed that the attribute separation unit 103 can execute the attribute separation algorithm capable of high-speed processing described with reference to FIG. If a buffer means for temporarily storing the read document image in the scanner 101 is provided, an attribute separating unit 103 that cannot perform real-time processing on the document scanning time can be used.
[0062]
The scanner 101 is connected to a personal computer 105 via a network cable 104, and transmits a signal for controlling the operation of the scanner 101 and a signal indicating the internal state of the scanner 101 to the personal computer 105 via the network cable 104. And the image data plane output from the scanner 101 and separated by attributes can be taken into the personal computer 105.
[0063]
The personal computer 105 has a disk drive to which a portable storage medium such as a CD-ROM 106 or a flexible disk 107 can be attached and detached, and stores a software program stored in the portable storage medium into the personal computer 105 via the disk drive. Can be read into memory.
[0064]
FIG. 11 is a diagram illustrating a personal computer. The personal computer 105 includes a display 111, a keyboard 112, and a mouse 113 as a pointing device. An image to be processed is displayed on the display 111, and by operating the mouse 113, it is possible to specify a region to be subjected to advanced attribute separation processing. This fulfills the function of the area designation section in the above-described embodiment.
[0065]
The display 111 includes a display image 115 corresponding to the original image data read by the scanner 101, a display image 117 corresponding to the selected data plane, and a background image data plane based on the multilayer image data plane output by the scanner 101. The corresponding display image 118 is displayed. The image corresponding to the foreground image data plane may be displayed instead of the display image corresponding to the original image, or the display image 118 corresponding to the background image data plane may not be displayed.
[0066]
In FIG. 11, the display image 118 corresponding to the background image data plane includes a portion where characters (but not black characters) and a pattern are superimposed, and the display image 117 corresponding to the selected data plane includes This illustrates that the character of the portion is not extracted. In this embodiment, in order to separate a character / line drawing part that cannot be separated by the separation processing of the attribute separation unit 103, the part that cannot be separated is executed by a personal computer. To perform the separation process again. For this re-separation process, the operator designates a target closed area with reference to the display image 115 of the original image data displayed on the display 111. As a method of designating a desired closed area with reference to the display 111, the operator operates the mouse 113, and for example, positions the mouse pointer 114 on the start point (upper left point of the rectangular area) of the area 116 where the attribute separation result is to be corrected. An operation method such as pressing the left button of the mouse 113 and operating the mouse 113 to the end point of the area 116 (the lower right point of the rectangular area) and dragging the mouse pointer 114 is possible. Note that this region designation operation is merely an example, and it is clear that region designation can be performed by other operations.
[0067]
The operation of the image processing apparatus shown in FIG. 10 will be described with reference to the flowchart in FIG. First, the operator first sets a document whose attribute is to be separated into a character / line drawing portion and a picture portion on the platen of the scanner 101 (S1201). Next, the operator issues a scan start command to the scanner 103 from the personal computer 105 to activate the scanner (S1202). In response to the scanning start command, the scanner 101 optically scans the original on the platen to generate digital image data corresponding to the original, and inputs the digital image data to the attribute separating unit 103. Here, if necessary, the image data input to the attribute separating unit 103 is converted into a color space suitable for the attribute separating process in the separating unit 103. The attribute separating unit 103 performs the first attribute separation on the input image data by detecting a black character portion included in the document by, for example, the algorithm shown in FIG. First multilayer image data including a plane, a background image data plane, and a selected data plane is generated (S1204). The personal computer 105 does not need to particularly control the first attribute separation processing in the scanner 101, and thus monitors the completion of the attribute separation processing in the scanner 101 (S1203). When the scanner 101 detects that the attribute separation processing has been completed, the scanner 101 loads the first multilayer image data into the personal computer 105 via the network cable 104 (S1205).
[0068]
The personal computer 105 generates display images 115, 117, and 118 to be displayed on the display 111 based on the captured first multilayer image data so that the operator can confirm the result of the attribute separation performed by the scanner 101. (S1206). Here, the display image 115 indicating the original document image can be generated by combining the character color data plane and the background image data plane according to the logic of the selection data of the first multilayer image data. In addition, each plane of the multi-layer image data plane can be displayed as it is, and in that case, it is not necessary to perform the combining process.
[0069]
Next, in order to display the separation result on a display, the resolution of the selected data plane and the background image data plane of the first multilayer image data and the resolution of the synthesized first multilayer image data are adjusted to match the display resolution. Conversion processing is performed (S1207). Each image whose resolution has been converted is displayed on a display as a first separation result, as illustrated in FIG. 11 (S1208).
[0070]
The operator refers to the first separation result displayed on the display, corrects the separation result, and designates a portion where attribute separation is to be performed more accurately as a closed area (S1209). At this time, a plurality of closed areas can be specified. The personal computer 105 cuts out only the image data in the closed area designated in this manner from the synthesized image data (corresponding to the image data when the document is read) (S1210). Then, the personal computer 105 executes a second attribute separation process based on the local property of the image according to the flowchart shown in FIG. 5 on the image data in the cut-out closed area (S1211), and based on the attribute separation result. The second multilayer image data is generated (S1212). As described with reference to FIG. 7, since the selection data included in the second multilayer image data can be combined with the selection data included in the first multilayer image data, the selection data is generated from two types of attribute separation algorithms. These two selection data are combined into one selection data (S1213). Next, the combined multilayer image data generated in step S1206 is attribute-separated using the combined selection data to generate a background image data plane and a character color data plane (S1214). This separation result is called a second separation result. When the second separation result is created, resolution conversion is performed on the second separation result as needed, and the result is displayed on the display in order to allow the operator to determine whether the result is good or not (S1215).
[0071]
The operator determines whether or not the desired attribute separation processing has been performed by viewing the result (S1216). If the result is not satisfied, the operator returns to step S1209 and repeats the operation from the designation of the area again. If the result is satisfied, the second separation result is stored as a final attribute separation result (S1217).
[0072]
It should be noted that the quality of the second separation result can be confirmed to a considerable extent only by displaying the selected data plane. Therefore, in step S1215, only the selected data plane may be displayed. In this case, it is not necessary to perform attribute separation at the stage of step S1214, and the attribute separation may be performed after determining that the selected data synthesized in step S1213 is good.
[0073]
The preferred embodiment of the present invention has been described above. In the above embodiment, the case where the attribute separation processing is performed by appropriately allocating two different types of attribute separation processing methods has been described. However, even when the number of types of the attribute separation processing methods is three or more, the present invention Applicable. In this case, the operator may separately designate which attribute separation process is to be assigned to each area.
[0074]
Each device configuration of FIGS. 1, 7, 9 and the like described above can be realized by software or can be configured by a hardware circuit. In the above description, the method shown in FIGS. 4 and 7 is exemplified as a plurality of attribute separation methods having different processing capacities. However, the present invention is not limited to this, and various methods can be adopted.
[0075]
As described above, according to the present embodiment, a closed region is specified for image data to be subjected to attribute separation, and for each closed region, for example, the attribute separation is not so high but the processing speed is high. Since a processing method and an attribute separation processing method having a low processing speed but a high separation performance can be assigned, it is possible to execute the attribute separation processing with high accuracy while suppressing a reduction in the processing speed of the entire image.
[Brief description of the drawings]
FIG. 1 is a diagram showing a schematic configuration of an image processing apparatus according to the present invention.
FIG. 2 is a diagram illustrating a configuration of an attribute separating unit.
FIG. 3 is a diagram showing an example of image data in which a character / line drawing portion and a picture portion are mixed.
FIG. 4 is a flowchart illustrating a simple attribute separation algorithm.
FIG. 5 is a flowchart illustrating a complicated attribute separation algorithm.
FIG. 6 is a diagram illustrating a result of attribute separation performed by applying an algorithm for extracting a shape of a character / line drawing portion.
FIG. 7 is a diagram for explaining another embodiment of the image processing apparatus according to the present invention.
FIG. 8 is a diagram illustrating an MRC having three or more layers.
FIG. 9 is a diagram illustrating another embodiment of the present invention that can generate an MRC having three or more layers.
FIG. 10 is a diagram illustrating another embodiment of the present invention.
FIG. 11 is a diagram illustrating a personal computer.
FIG. 12 is a flowchart illustrating an example of a processing procedure in a configuration using a personal computer.
[Explanation of symbols]
1 image data input section, 2 area designation section, 3 switching section, 4 first attribute separation section, 5 second attribute separation section.

Claims (9)

A plurality of attribute separating means for separating image data into a character / line drawing portion and a picture portion by different methods,
Control means for receiving designation of area division for the image to be processed, and causing the corresponding attribute separation means to process data of each area of the image to be processed according to the designation of area division;
An image processing apparatus comprising: The image data is a first foreground image data, a first background image data, and a first pixel for each pixel indicating which one of the first foreground image data and the first background image data is to be selected. First separation means for separating the data into three layers, namely, one selection data;
The image data is converted into a second foreground image data, a second background image data, and the second foreground image data and the second background image data by a method different from the first separation means. Second selection data indicating which one to select, and second separation means for separating into three layers:
A control unit that receives designation of area division for an image to be processed, and distributes and processes data of each area of the image to be processed to the first separation unit and the second separation unit in accordance with the region division designation When,
An image processing apparatus comprising: 3. The image processing apparatus according to claim 2, wherein the first selection data generated by the first separation unit is combined with the second selection data generated by the second separation unit. An image processing apparatus comprising: a synthesizing unit configured to generate an image. The third selection data is generated based on the area specifying information output by the area specifying means, and the three-layer data output by the first separating means and the three-layer data output by the second separating means are output. 3. The image processing apparatus according to claim 2, further comprising: a multi-layer image data generation unit configured to generate a multi-layer image data plane including the data and the third selection data. First attribute separating means for separating image data into a character / line drawing portion and a picture portion;
Display means for displaying information related to the separation result output by the first attribute separation means;
Area designation means for receiving designation of a closed area on the image data in association with the display of the display means,
A second attribute separating unit that separates the image data included in the closed region specified by the region specifying unit into a character / line drawing portion and a picture portion in a different manner from the first attribute separating unit;
An image processing apparatus comprising: The image data is a first foreground image data, a first background image data, and a first pixel for each pixel indicating which one of the first foreground image data and the first background image data is to be selected. A first separation step of separating the selected data into three layers;
The image data is converted into a second foreground image data, a second background image data, and the second foreground image data and the second background image data by a method different from the first separation step. Second selection data indicating which to select, a second separation step of separating into three layers,
Receiving designation of area division for the image to be processed, and selectively executing the first separation step or the second separation step on data of each area of the image to be processed in accordance with the designation of area division. A control step;
An image processing method including: A first attribute separation step of separating the image data into a character / line drawing part and a picture part;
A display step of displaying information related to a separation result of the first attribute separation step;
Area designation means for accepting designation of a closed area on the image data in association with the display of the display step;
A second attribute separating step of separating the image data included in the closed area specified by the area specifying means into a character / line drawing portion and a picture portion in a manner different from the first attribute separating step;
An image processing method including: The image data is a first foreground image data, a first background image data, and a first pixel for each pixel indicating which one of the first foreground image data and the first background image data is to be selected. A first separation unit for separating the data into three layers:
The image data is converted into a second foreground image data, a second background image data, and the second foreground image data and the second background image data by a method different from the first separation means. Second selection data indicating which one to select, and second separation means for separating into three layers:
A control unit that receives designation of region division for an image to be processed, and distributes and processes data of each region of the image to be processed to the first separation unit and the second separation unit in accordance with the region division designation ,
As a program for operating a computer system. First attribute separating means for separating image data into a character / line drawing portion and a picture portion;
Display means for generating display data indicating information related to the separation result output by the first attribute separation means;
Area specifying means for receiving designation of a closed area on the image data in association with the display data generated by the display means;
A second attribute separating unit that separates the image data included in the closed region specified by the region specifying unit into a character / line drawing portion and a picture portion by a method different from the first attribute separating unit;
As a program for operating a computer system.

Images