| 摘要 |
A technique for automatically extracting a region of interest from a set of a plurality of images includes a scheme exploiting a motion vector, and a scheme based on an active contour model. But these schemes suffer various problems (e.g., detection resolution and precision are low, resolutions in units of pixels cannot be obtained, versatility is poor, and so forth). To solve these problems, when initial contour information in an image sensed at reference time is input, a feature point selector selects a plurality of feature points from its contour line, and stores the connectivity between these feature points in a memory. A corresponding point seeker seeks a plurality of corresponding points, which respectively correspond to the plurality of selected feature points, from an image to be sought, which is sensed at another time, and the contour between the plurality of found corresponding points is extracted on the basis of the stored connectivity.
|
