JP4766661B2 - Image processing method and image processing apparatus - Google Patents

Description

  The present invention relates to an image processing method and an image processing apparatus that divide an image obtained by scanning a paper document and vectorize the image for each region.

  In recent years, the digitization of information has progressed, and systems that do not store paper documents as they are but instead store them electronically or transmit the electronic data to other devices have become widespread. Further, the document to be digitized is not limited to a monochrome binary image but is expanding to a full-color (multi-value) image electronic document.

  Further, the target electronic document is not limited to a document obtained by scanning a document on paper with a scanner or the like into image data. For example, an electronic document is obtained by separating a region of a document image, performing character recognition processing on a character region and converting it into a character code string, converting a photo region into vector data of a contour line, etc. It includes a document image that has been converted into higher information and generated (see, for example, Patent Document 1).

  Among the vectorized images as described above, even a full-color image includes an image obtained by scanning an original document created by an illustration or graphic creation software and vectorizing it. These images have features such as clear outlines of objects and limited appearance colors compared to natural images such as photographs. Hereinafter, such an image is referred to as a “clip art image”.

In the road data generation method disclosed in Patent Document 2, first, a photographic image input in full color is converted into a binary image. Next, the contour line and the center line are extracted from the binary image, and the color information of the obtained line and the original image is converted into vector data. The noise processing is described as removing isolated noise by performing expansion / contraction processing.
JP 2004-265384 A JP 2004-246554 A

  However, in the conventional processing as described above, a useful area may be mistakenly deleted as a noise area. In such a case, an accurate contour line cannot be obtained, and the image after vectorization has a picture quality. It will be deteriorated. In addition, when the contour line is vectorized while leaving the noise region without performing noise removal, there is a problem that the amount of vector data becomes enormous.

  The present invention has been made in view of such circumstances, and can suitably generate vector data in which the amount of data is reduced by reducing noise for a graphic region selected from a document image ( In particular, an object of the present invention is to provide an image processing method and an image processing apparatus capable of generating suitable vector data for a clip art image area such as an illustration.

In order to solve the above problems, an image processing method according to the present invention includes:
A clustering step of clustering the image to be vectorized on the basis of the similarity of the color features to generate a plurality of clusters in the image ;
A determination step of determining, as noise , a region whose size is smaller than a noise determination threshold among a plurality of regions included in each cluster generated in the image ;
Removal means, said noise and the determined area determination step, in the other clusters adjacent to each other on the image in the area, it is integrated into cluster similarity of the color features are determined and the highest To remove the region determined to be the noise,
The converting means includes a converting step of converting a cluster from which the area determined to be the noise in the removing step is removed into vector data.

  According to the present invention, it is possible to suitably generate vector data in which the amount of data is reduced by reducing noise for a graphic region selected from a document image.

  Hereinafter, a vectorization process of image data using an image processing apparatus according to an embodiment of the present invention will be described in detail with reference to the drawings.

<First Embodiment>
[Device configuration]
FIG. 1 is a block diagram illustrating a configuration of an image processing apparatus having a function of performing vectorization processing based on region division according to the first embodiment of the present invention. In FIG. 1, an input unit 11 inputs a color document image by scanning a paper document or the like. The area separating unit 12 separates the color document image input by the input unit 11 into a plurality of types of areas including a photographic area. Furthermore, the clip art image selection unit 13 selects a clip art image from the regions separated by the processing of the region separation unit 12.

  Furthermore, in FIG. 1, the area dividing unit 14 divides the clip art image selected by the clip art image selecting unit 13 into a plurality of areas based on the color characteristics. Furthermore, the region integration unit 15 integrates similar regions among the plurality of divided regions. Furthermore, the noise determination unit 16 determines a noise region from each of the divided and integrated regions. Furthermore, the noise removing unit 17 removes the area determined as the noise area by the noise determining unit 16. Then, the vectorization unit 18 converts the regions obtained as a result of the region division (regions after integration when integrated, regions after noise removal when noise is removed) into vector data, respectively.

  FIG. 10 is a block diagram showing a main part configuration of a digital multi-function peripheral (MFP) which is an embodiment realizing the image processing apparatus shown in FIG. In this embodiment, a digital multifunction peripheral (MFP) having a scanner function and a printer function is used as the image processing apparatus. However, a system in which a general-purpose scanner and a personal computer are connected is used as the image processing apparatus. Also good.

  As shown in FIG. 10, the MFP includes a controller unit 2000 that functions as an image processing apparatus. The controller unit 2000 connects a scanner 2070 as an image input device and a printer 2095 as an image output device, and realizes a copy function for printing out image data read from an original image by the scanner 2070 by the printer 2095. Take control. Further, the controller unit 2000 performs control for inputting / outputting pattern images, device information, and the like to / from other apparatuses via the LAN 1006 and the public line (WAN) 1008.

  As shown in FIG. 10, the controller unit 2000 has a CPU 2001. The CPU 2001 starts up an operation system (OS) by a boot program stored in the ROM 2003, and executes various processes by executing application programs stored in an HDD (hard disk drive) 2004 on the OS. A RAM 2002 is used as a work area for the CPU 2001. The RAM 2002 provides not only a work area for the CPU 2001 but also an image memory area for temporarily storing image data. The HDD 2004 stores image data together with the application program.

  Along with the ROM 2003 and the RAM 2002, the CPU 2001 has an operation unit I / F (operation unit interface) 2006, a network I / F (network interface) 2010, a modem 2050, and an image bus I / F (image bus interface). ) 2005 is connected.

  An operation unit I / F 2006 is an interface with an operation unit 2012 having a touch panel, and outputs image data to be displayed on the operation unit 2012 to the operation unit 2012. Further, the operation unit I / F 2006 sends information input by the user through the operation unit 2012 to the CPU 2001.

  The network I / F 2010 is connected to the LAN 1006 and inputs / outputs information to / from each device connected to the LAN 1006 via the LAN 1006. The modem 2050 is connected to the public line 1008 and inputs / outputs information to / from other devices via the public line 1008.

  An image bus I / F 2005 is a bus bridge for connecting a system bus 2007 and an image bus 2008 for transferring image data at high speed and converting a data structure. The image bus 2008 includes a PCI bus or IEEE1394. On the image bus 2008, a raster image processor (RIP) 2060, a device I / F 2020, a scanner image processing unit 2080, a printer image processing unit 2090, an image rotation unit 2030, a thumbnail creation unit 2035, and an image compression unit 2040 are provided. Yes.

  The RIP 2060 is a processor that develops a PDL code into a bitmap image. A scanner 2070 and a printer 2095 are connected to the device I / F 2020 and perform synchronous / asynchronous conversion of image data. A scanner image processing unit 2080 performs correction, processing, and editing processing on input image data. A printer image processing unit 2090 performs printer correction, resolution conversion, and the like on print output image data. The image rotation unit 2030 rotates image data. The image compression unit 2040 compresses the multi-value image data into JPEG data and the binary image data into data such as JBIG, MMR, and MH, and also performs decompression processing.

[Outline of vectorization processing based on area division]
FIG. 2 is a flowchart for explaining details of the vectorization process based on the area division of the clipart image according to the first embodiment of the present invention.

  In the image processing apparatus, first, color document image data is obtained by scanning a paper document in the input unit 11 (step S11). Next, the area separation unit 12 performs binarization processing on the color document image input in step S11 to convert it to binary image data, and the binary image data is converted into a plurality of characters, photographs, tables, and the like. Separated into types of regions (step S12). As an example for realizing this region separation processing, a technique described in US Pat. No. 5,680,478 “Method and Apparatus for character recognition (Shin-YwanWang et al./Canon KK)”. Can be used. The region separation processing in this embodiment also uses the technology.

  Next, the clip art image selection unit 13 selects a clip art image from the regions separated in the region separation process of step S12 (step S13). Further, the area dividing unit 14 performs clustering processing based on the color feature of each selected clipart image, and divides the clip art image into clusters (area division) (step S14). The details of this area division processing will be described later.

  Then, the region integration unit 15 performs region integration by integrating the clusters divided in the step S14 based on the similarity (step S15). Details of this area integration processing will be described later.

  After that, the result of area division is labeled in the noise determination unit 16, the outline of each cluster is traced to count the pixels constituting the outline (step S16), and the outline of each label area is constituted. By comparing the size of the number of pixels with a certain threshold value, it is determined whether or not the region is a noise region (step S17). As a result, a label area that is somewhat small is determined as a noise area (Yes). On the other hand, if the label area is larger than a certain size, it is not determined as a noise area (No), and the process proceeds to step S19. Details of the noise determination process will be described later.

  Then, the noise removal unit 17 performs the noise removal process by performing the clustering process again on the noise pixels included in the noise area determined in step S17 based on the similarity between the adjacent areas (step S18). ), And proceeds to step S19. Details of the noise removal process will be described later.

  In step S19, it is determined whether or not processing has been completed for all label regions, that is, whether or not processing has been completed for all label regions. As a result, when the processing target still exists (No), the process returns to step S17, and the above-described noise region determination processing (step S17) and noise region removal processing (step S18) are repeatedly executed for the region. On the other hand, when the processing of all the label areas has already been completed and there is no processing target (Yes), the process proceeds to step S20.

  In step S20, the vectorization unit 18 converts each divided area into vector data based on the outline and the color inside the area. As a technique for realizing this vectorization processing, for example, there is a technique of vectorizing by tracing the contour line of a binary image of Japanese Patent No. 2885999 and selecting the coordinate vector. Note that the technique is also used for the vectorization processing according to the present embodiment.

[Clip art image selection example]
FIG. 3 is a diagram illustrating an example in which a clip art image is selected from a document image in the image processing according to the first embodiment of the present invention. FIG. 3 shows a state in which a photographic area 31, a text area 32, and a clipart area 33 are separated as rectangular areas from one document image by using the area separation method described above.

[Area division processing]
Hereinafter, the region division processing (step S14) in the flowchart shown in FIG. 2 will be described in detail. FIG. 4 is a flowchart for explaining in detail the area dividing process (step S14) in the flowchart shown in FIG.

  First, the first cluster is generated by the raster-scanned start pixel (step S1401). Next, the similarity between all the clusters is obtained for the next pixel (step S1402). The higher the similarity, the closer the characteristics of the pixel and cluster. In the similarity calculation process, for example, RGB color information is used, but other color space information or information other than color may be used as the feature amount.

  Then, the highest similarity and the cluster number corresponding to this similarity are recorded, and this similarity is compared with a preset threshold (step S1403). If the result is higher than the threshold (Yes), the target pixel belongs to the recorded cluster (step S1404). On the other hand, if it is lower than the threshold (No), a new cluster is generated for the target pixel (step S1405). When the similarity between the clusters with respect to the pixel is obtained by the inter-pixel distance, the case where the similarity is higher than the threshold means that the inter-pixel distance is smaller than the threshold.

  After the process of step S1404 or step S1405, it is determined whether or not the process for all the pixels has been completed (step S1406). As a result, when there are still unprocessed pixels (No), the process returns to step S1402, and the above-described processing is repeatedly executed. On the other hand, if there is no unprocessed pixel (Yes), the area division process (step S14) is terminated.

[Area integration processing]
Next, the region integration processing (step S15) in the flowchart shown in FIG. 2 will be described in detail. FIG. 5 is a flowchart for explaining in detail the region integration processing (step S15) in the flowchart shown in FIG.

  First, a target value for the number of regions to be finally separated is set (input) (step S1501). In the present embodiment, as an example, it is assumed that how many colors are separated as a guide. Next, the number of current clusters is counted (step S1502). Then, the counted current number of clusters is compared with the set target value (step S1503).

  As a result, when the current number of clusters is greater than the target value (Yes), cluster integration is performed. In the present embodiment, as cluster integration processing, first, the similarity (distance) between the clusters is calculated (step S1504), and the two clusters with the highest similarity (two clusters with the shortest distance) are calculated. ) To be integrated, and the two clusters to be integrated are integrated into one cluster (step S1505).

  After the above-described cluster integration process is completed, the process returns to step S1502 and the above-described process is repeatedly performed on the cluster group after the integration process. In the comparison process in step S1503, when the current number of clusters is smaller than the set target value (No), the area integration process (step S15) is terminated.

[Noise judgment processing]
Next, the noise determination process (step S17) in the flowchart shown in FIG. 2 will be described in detail with reference to FIG. FIG. 6 is a diagram for explaining in detail the noise determination process (step S17) in the flowchart shown in FIG.

  In FIG. 6, a cluster 61 and a cluster 62 are two examples of clusters selected as representatives from the clusters after the region division processing (step S14) and the region integration processing (step S15). These clusters 61 and 62 shown in FIG. 6 contain many small areas, and if these are converted as they are into the vector data of the contour lines of the clusters and the internal color information, the amount of data becomes enormous. It becomes a problem. Therefore, in order to solve this problem, the result of area division is labeled as described above (step S16), and it is determined whether the area is a noise area based on the size of each label area (step S17). . As a result, each area included in the area 63 is determined as a noise area, and each area is a target of noise removal processing.

[Noise removal processing]
Next, the noise removal process (step S18) in the flowchart shown in FIG. 2 will be described in detail. FIG. 7 is a flowchart for explaining in detail the noise removal processing (step S18) in the flowchart shown in FIG. In the noise removal process, the noise area determined in the noise determination process (step S17) is set as a noise removal process target, and the removal process is performed for each noise pixel included in the noise area.

  Therefore, first, the similarity between the noise pixel and each adjacent cluster (for example, the distance between the noise pixel and the adjacent region) is calculated (step S1801). Next, the noise pixel is assigned to the cluster having the highest similarity from the calculated similarity (integrated with the adjacent cluster having the highest similarity), and the region to which the noise pixel belongs is updated (step S1802). Then, it is determined whether or not an unprocessed noise pixel exists in this noise region (step S1803).

  As a result, when there is an unprocessed pixel (Yes), the process returns to step S1801 to repeat the above process. On the other hand, when there is no unprocessed pixel (No), the noise removal processing of this noise region is finished.

[Example of vectorization based on segmentation of clip art image]
FIG. 8 is a diagram illustrating an example of vectorization based on region division of a clip art image according to an embodiment of the present invention. In FIG. 8, 81 is a clip art image divided into regions. Reference numeral 82 denotes an example of a region division result. That is, the clip art image 81 is displayed in the region division result 82 when the target value of the number of regions to be divided is set to 16, for example, by the above-described series of processing of region division, region integration, noise determination, and noise removal. The cluster is divided into 16 clusters as shown.

  As examples of cluster outlines and internal color information necessary for vectorization processing, FIG. 8 shows a cluster 83, outline 84, and internal color information 85. A result of converting the region division result 82 into vector data based on the above information is a vector image 86. Since the file size of the vector image 86 is 24 Kbytes, and the file size of the original clip art image 81 is 2739 Kbytes, it is reduced to 1/100 or less by vectorization. Further, processing such as painting of the internal color of each cluster can be performed using each cluster 82 of the vector image 86 as a component.

  As described above, according to the present embodiment, noise can be effectively removed by re-clustering the noise region determined by the labeling process. Also, the amount of data can be reduced by vectorizing the result of the area division from which noise is removed.

<Second Embodiment>
Hereinafter, a second embodiment according to the present invention will be described.

  In the first embodiment described above, an example has been described in which it is determined whether or not the area is a noise area based on the size of the label area after the labeling process. In the second embodiment, in order to prevent useful information from being removed from the image, after determining the noise area based on the size of the label area, it is further determined whether the area is a noise area using other information. An example will be described with reference to FIG. Note that the configuration of the image processing apparatus having a function of performing vectorization processing based on region division in the second embodiment is the same as that of the image processing apparatus according to the first embodiment described above.

  FIG. 9 is a flowchart for explaining details of the vectorization process based on the area division of the clipart image according to the second embodiment of the present invention. In the flowchart of FIG. 9, steps S1701 and S1702 are noise region determination portions in the second embodiment, and other steps are the same as those in the first embodiment described above.

  First, in step S1701, it is determined whether or not it is a noise area candidate based on the size of the label area. As a result, if the label area is not small (No), it is determined that the label area is not a noise area, and the process proceeds to step S19. On the other hand, if it is determined that the label area is small (Yes), it is recorded as a noise area candidate, and the process proceeds to step S1702.

  Next, in step S1702, it is determined whether the noise area recorded in step S1701 is a noise area depending on whether there is an edge around it. As a result, when there is an edge pixel around the noise area candidate (Yes), it is determined that the area is a noise area, and the process proceeds to the noise removal process in step S18. On the other hand, when there is no strong edge pixel around the noise area candidate (No), it is determined that the area is not a noise area, and the process proceeds to step S19. Note that an edge pixel indicates a pixel whose color or luminance changes greatly from an adjacent pixel. For example, it can be determined whether or not the pixel is an edge by using a Sobel filter. Here, it is determined whether or not the noise is generated by paying attention to the characteristics of the scanner that the noise easily occurs around the edge and the characteristics such as JPEG compression.

  In addition to the edge information, the noise region may be determined based on the positional relationship with the surrounding region with respect to the noise region determined based on the size of the label region. That is, if there is a large area of the same color around a small label area, the area is reliably determined as a noise area.

  As described above with reference to the first and second embodiments, according to the present invention, it is possible to improve the image quality by accurately taking the outline of each area by performing the area division process from which noise is removed. It is possible to reduce the amount of data for vectorization description of each region outline and obtain a good image component.

<Other embodiments>
Although the embodiments have been described in detail above, the present invention can take embodiments as, for example, a system, an apparatus, a method, a program, or a storage medium (recording medium). The present invention may be applied to a system composed of a single device or an apparatus composed of a single device.

  In the present invention, a software program (in the embodiment, a program corresponding to the flowchart shown in the figure) that realizes the functions of the above-described embodiment is directly or remotely supplied to the system or apparatus, and the computer of the system or apparatus Is also achieved by reading and executing the supplied program code.

  Accordingly, since the functions of the present invention are implemented by computer, the program code installed in the computer also implements the present invention. In other words, the present invention includes a computer program itself for realizing the functional processing of the present invention.

  In that case, as long as it has the function of a program, it may be in the form of object code, a program executed by an interpreter, script data supplied to the OS, or the like.

  As a recording medium for supplying the program, for example, floppy (registered trademark) disk, hard disk, optical disk, magneto-optical disk, MO, CD-ROM, CD-R, CD-RW, magnetic tape, nonvolatile memory card ROM, DVD (DVD-ROM, DVD-R) and the like.

  As another program supply method, a client computer browser is used to connect to an Internet homepage, and the computer program of the present invention itself or a compressed file including an automatic installation function is downloaded from the homepage to a recording medium such as a hard disk. Can also be supplied. It can also be realized by dividing the program code constituting the program of the present invention into a plurality of files and downloading each file from a different homepage. That is, a WWW server that allows a plurality of users to download a program file for realizing the functional processing of the present invention on a computer is also included in the present invention.

  In addition, the program of the present invention is encrypted, stored in a storage medium such as a CD-ROM, distributed to users, and key information for decryption is downloaded from a homepage via the Internet to users who have cleared predetermined conditions. It is also possible to execute the encrypted program by using the key information and install the program on a computer.

  In addition to the functions of the above-described embodiments being realized by the computer executing the read program, the OS running on the computer based on the instruction of the program is a part of the actual processing. Alternatively, the functions of the above-described embodiment can be realized by performing all of them and performing the processing.

  Furthermore, after the program read from the recording medium is written in a memory provided in a function expansion board inserted into the computer or a function expansion unit connected to the computer, the function expansion board or The CPU or the like provided in the function expansion unit performs part or all of the actual processing, and the functions of the above-described embodiments are realized by the processing.

It is a block diagram which shows the structure of the image processing apparatus which has a function which performs a vectorization process based on the area | region division based on the 1st Embodiment of this invention. It is a flowchart for demonstrating the detail of the vectorization process based on the area | region division | segmentation of the clipart image which concerns on the 1st Embodiment of this invention. It is a figure which shows an example which selected the clip art image from the document image in the image processing which concerns on the 1st Embodiment of this invention. FIG. 3 is a flowchart for explaining in detail an area division process (step S14) in the flowchart shown in FIG. 2; FIG. FIG. 3 is a flowchart for explaining in detail an area integration process (step S15) in the flowchart shown in FIG. 2; It is a figure for demonstrating in detail the noise determination process (step S17) in the flowchart shown in FIG. 3 is a flowchart for explaining in detail a noise removal process (step S18) in the flowchart shown in FIG. It is a figure which shows an example of vectorization based on the area | region division of the clip art image by one Embodiment of this invention. It is a flowchart for demonstrating the detail of the vectorization process based on the area | region division | segmentation of the clip art image which concerns on the 2nd Embodiment of this invention. FIG. 2 is a block diagram illustrating a main part configuration of a digital multifunction peripheral (MFP) that is an embodiment that realizes the image processing apparatus illustrated in FIG. 1.

Explanation of symbols

DESCRIPTION OF SYMBOLS 11 Input part 12 Area separation part 13 Clip art image selection part 14 Area division part 14 Area integration part 16 Noise determination part 17 Noise removal part 18 Vectorization part

Claims (11)

A clustering step of clustering the image to be vectorized on the basis of the similarity of the color features to generate a plurality of clusters in the image ;
A determination step of determining, as noise , a region whose size is smaller than a noise determination threshold among a plurality of regions included in each cluster generated in the image ;
Removal means, said noise and the determined area determination step, in the other clusters adjacent to each other on the image in the area, it is integrated into cluster similarity of the color features are determined and the highest To remove the region determined to be the noise,
An image processing method comprising: a conversion step of converting a cluster from which the area determined to be the noise in the removal step is removed into vector data. In the clustering step, the image to be vectorized is clustered based on the similarity of the color features to generate a plurality of clusters in the image, and the number of the generated clusters is a target value. If the number is larger, the plurality of clusters having similar color characteristics among the plurality of generated clusters are integrated, and the number of clusters is updated until the target value is reached. The image processing method according to claim 1. The setting means further comprises a setting step of setting the target value of the number of clusters to be generated;
The image processing method according to claim 2, wherein in the clustering step, the clusters are updated until the number of the generated clusters reaches the target value. In the determination step, if an edge pixel is present around the area where the size of the area is determined to be smaller than the noise determination threshold, the area is determined as noise. The image processing method according to claim 1 . Wherein in the determination step, relative to the size of the region is determined to be smaller than the noise determination threshold region, further, has a size and of more than the noise judgment threshold have a same color characteristics around the area The image processing method according to claim 1 , wherein the area is determined as noise on condition that another area exists. A dividing step of dividing the document image into a plurality of types of regions;
A selection step further comprising: a selection step of selecting an image region having a number of colors smaller than a predetermined number from a plurality of types of regions divided in the division step;
The image processing method according to any one of claims 1 to 5, wherein the vector data processing target image is an image region selected in the selection step. The vector data processing target image is an electronic image obtained by reading a document on which an illustration created using an illustration creation application is printed, or an illustration drawn on a paper manuscript. The image processing method according to claim 1, wherein the image processing method is a region formed of a digitized image. In the determination step, a plurality of regions included in each cluster are determined by performing a labeling process on each cluster, and the size of the region among the plurality of regions determined by the labeling process is The image processing method according to claim 1, wherein an area smaller than a noise determination threshold is determined as the noise. Clustering means for generating a plurality of clusters in the image by clustering the image to be vectorized based on the similarity of the color features;
A determination unit that determines, as noise , a region whose size is smaller than a noise determination threshold among a plurality of regions included in each of the generated clusters;
Said region determined as noise determining means, among other clusters adjacent to each other on the image in the area, by integrating the cluster similarity of the color features are determined and the highest, the noise Removing means for removing the region determined to be,
An image processing apparatus comprising: a conversion unit configured to convert a cluster from which the area determined to be the noise by the removing unit is removed to vector data. Computer
Clustering means for generating a plurality of clusters in the image by clustering the image to be vectorized based on the similarity of the color features,
A determination unit that determines, as noise , a region having a size smaller than a noise determination threshold among a plurality of regions included in each of the generated clusters.
Said region determined as noise determining means, among other clusters adjacent to each other on the image in the area, by integrating the cluster similarity of the color features are determined and the highest, the noise Removing means for removing the region determined to be,
Conversion means for converting the cluster from which the area determined to be the noise by the removal means is removed into vector data;
Program to function as. A computer-readable storage medium storing the program according to claim 10 .

Images