| 摘要 |
PROBLEM TO BE SOLVED: To provide an image analyzing device capable of extracting a document area from an input image with higher accuracy. SOLUTION: A document edge rough detection process is performed for an input image to detect document edges roughly, and a document edge considered for the certainty of a detection result to be high is identified (S20). Next, using the roughly detected document edge information, a underlying color candidate and a background color candidate are selected respectively out of a pixel value distribution around center and a pixel value distribution around outside of the document edges (S30). Then, a pixel having the underlying color candidate around center of the image and the background color candidate around outside of the image is detected, and the detected pixel is selected as a document edge pixel (S40). Finally, the final document edges are generated from the input image by replacing the selected image edge pixels with a straight line (S50). COPYRIGHT: (C)2004,JPO
|
