| 摘要 |
PROBLEM TO BE SOLVED: To accurately detect an edge component of a target even when a source image has much noise on its background part by extracting feature parameters from a linear distribution diagram generated by integrating respective generated edge images in one direction. SOLUTION: The source image is read out and a frequency (n) of movement for shifting the source image in one direction is found from target size and luminance features of its image (S1). Then a movement frequency counter k=1 is set (s2) and the source image is moved by a specific number of pixels to generate a shifted image (S3); and the luminance of the shifted image is compared with that of the source image to generate an edge image (S4). The edge image is moved and projected orthogonally to generate a linear distribution diagram (S5), and feature parameters of the edge component are extracted (S6). Then count-up operation is performed (S7) and while a movement frequency is judged (S8), an edge image when counter k=n is found; and the generation frequencies of the feature parameters are totalized and the edge component of the target is extracted on the basis of a specific evaluation function (S9 and S10). |
