IMAGE PROCESSING APPARATUS, IMAGE PROCESSING METHOD, AND PROGRAM

• US 20070086667A1
• Filed: 10/06/2006
• Published: 04/19/2007
• Est. Priority Date: 10/17/2005
• Status: Active Grant

• Current Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Original Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Inventors: DAI, XIAOYAN, Osawa, Hidefumi
• Granted Patent: US 7,831,107 B2
• Time in Patent Office: Days
• US Class Current: 382/242
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/15 Cutting or merging image elements, e.g. region growing, watershed or clustering-based techniques
• CPC Class Codes: G06V 30/18076 by analysing connectivity, e.g. edge linking, connected component analysis or slices
• CPC Class Codes: G06V 30/1823 using vector-coding

Abstract

An image processing apparatus, image processing method, and program which enable generation of vector data capable of replicating a clipart region of a raw image as faithfully as possible are provided. To accomplish this, a raw image is segmented into region images according to attributes. From the segmented region images, a region image having a predetermined attribute is cut out from the raw image. At least one representative color which configures the predetermined region image is decided based on an appearance color of the cut out predetermined region image. A color image having the decided representative color is extracted from the predetermined region image. A contour line of the extracted color image is extracted. An edge image within the predetermined region image is extracted. The extracted contour line is corrected based on the extracted edge image. Using the corrected contour line, vector data of the predetermined region image is generated.

Claims

1. An image processing apparatus adapted to execute image processing which converts a raw image into vector data, the image processing apparatus comprising:

a segmentation unit adapted to segment the raw image into region images for each attribute;

a cutout unit adapted to cut out region images segmented by the segmentation unit according to attribute from the raw image;

a determination unit adapted to determine at least one representative color configuring the region image cut out by the cutout unit, based on an appearance color of the region image;

an image extraction unit adapted to extract a color image having the representative color determined by the determination unit from the region image;

a contour line extraction unit adapted to extract a contour line of a color image extracted by the image extraction unit;

an edge image extraction unit adapted to extract an edge image within the region image;

a correction unit adapted to correct a contour line extracted by the contour line extraction unit based on an edge image extracted by the edge image extraction unit; and

a generation unit adapted to generate vector data of the region image using contour lines corrected by the correction unit.

2. The image processing unit of claim 1, wherein the generation unit generates contour line information, consisting of a contour line corrected by the correction unit, a color and width thereof, and a drawing instruction for drawing a color inside the contour line, as vector data of a region image having the attribute.

3. The image processing unit of claim 1, wherein the region image cut out by the cutout unit is a clipart region among region images segmented by the segmentation unit in which a number of appearance colors is below a threshold.

4. The image processing unit of claim 1, wherein the correction unit superimposes on the color image a corresponding edge image to generate a composite image, draws a region of the composite image surrounded by a contour line of the color image and a contour line of the edge image, in which both images do not overlap each other, with the color of the color image, and corrects the contour line of the composite image to the contour line of the edge image.

5. An image processing method for converting a raw image into vector data, the image processing method comprising the steps of:

segmenting the raw image into region images for each attribute;

cutting out region images segmented in the segmentation step according to attribute from the raw image;

determining at least one representative color configuring the region image cut out in the cutout step, based on an appearance color of the cut-out region image;

extracting a color image having the representative color determined in the determination step from the region image;

extracting a contour line of a color image extracted in the image extraction step;

extracting an edge image within the region image;

correcting a contour line extracted in the contour line extraction step based on an edge image extracted in the edge image extraction step; and

generating vector data of the region image using contour lines corrected in the correction step.

6. A computer-readable storage medium storing a computer program for controlling an image processing apparatus executing image processing for converting a raw image into vector data, the program causing a computer to execute the steps of:

segmenting the raw image into region images for each attribute;

cutting out region images segmented in the segmentation step according to attribute from the raw image;

determining at least one representative color configuring the region image cut out in the cutout step, based on an appearance color of the cut-out region image;

extracting a color image having the representative color determined in the determination step from the region image;

extracting a contour line of a color image extracted in the image extraction step;

extracting an edge image within the region image;

correcting a contour line extracted in the contour line extraction step based on an edge image extracted in the edge image extraction step; and

generating vector data of the region image using contour lines corrected in the correction step.

7. An image processing method for inputting a document image including a clipart image to an image processing apparatus to perform vectorization, the image processing method comprising the steps of:

region-segmenting the document image and selecting the clipart image;

extracting edge information from the clipart image;

forming an edge cluster based on the edge information;

segmenting a pixel portion of the clipart image other than the edge cluster into a plurality of clusters based on color characteristics;

fusing two or more clusters other than the edge cluster which are similar to each other; and

converting each cluster after region fusion to vector data based on a contour line and an internal color of each cluster.

8. The image processing method of claim 7, further comprising the steps of:

determining a new edge cluster from the plurality of clusters segmented in the segmentation step; and

fusing a newly determined edge cluster with an edge cluster formed in the edge cluster forming step.

9. The image processing method of claim 7, wherein the edge cluster forming step forms an edge cluster using an edge pixel having an edge strength that is higher than a threshold, based on edge information extracted in the extraction step.

10. The image processing method of claim 8, wherein the cluster determination step comprises the steps of:

calculating a edge ratio of a cluster other than the initially formed edge cluster, which contains edge information; and

determining the cluster as an edge cluster based on the magnitude of the edge ratio.

11. The image processing method of claim 7, further comprising the steps of:

obtaining label areas by labeling the plurality of clusters segmented in the segmentation step, and determining whether each label area is a noise area based on the size of the label area;

calculating a degree of similarity of each pixel within an noise area and other adjacent clusters; and

reprocessing the noise area so that each pixel within the noise area is fused with a cluster among the other clusters which has the highest degree of similarity.

12. An image processing apparatus adapted to vectorize a document image including a clipart image, the image processing apparatus comprising:

an input unit adapted to input the document image;

a selection unit adapted to region-segment the document image and select the clipart image;

an extraction unit adapted to extract edge information from the clipart image;

an edge cluster formation unit adapted to form an edge cluster based on the edge information;

a segmentation unit adapted to segment a pixel portion of the clipart image other than the edge cluster into a plurality of clusters based on color characteristics;

a region fusion unit adapted to fuse two or more clusters other than the edge cluster which are similar to each other; and

a conversion unit adapted to convert each cluster after region fusion to vector data based on a contour line and an internal color of each cluster.

13. A computer-readable storage medium storing a computer program for causing a computer, which inputs a document image including a clipart image to perform vectorization, to execute the procedures of:

region-segmenting the document image and selecting the clipart image;

extracting edge information from the clipart image;

forming an edge cluster based on the edge information;

segmenting a pixel portion of the clipart image other than the edge cluster into a plurality of clusters based on color characteristics;

fusing two or more clusters other than the edge cluster which are similar to each other; and

converting each cluster after region fusion to vector data based on a contour line and an internal color of each cluster.