초록 ▼

An image segmentation system uses deformable contours to automatically identify the border of a target object in each image of a set of images. The majority of contour deformations, or moves, are performed in O(1) time. Each contour uses both global and local geometric information and pixel intensit

An image segmentation system uses deformable contours to automatically identify the border of a target object in each image of a set of images. The majority of contour deformations, or moves, are performed in O(1) time. Each contour uses both global and local geometric information and pixel intensity information extracted from a training contour, as well as a priori domain knowledge, to find the outline of the target object.

대표청구항 ▼

[ What is claimed is:] [1.] A method of segmenting a set of images comprising the steps of:a) providing a set of search images, each search image containing a view of a target object;b) providing a priori domain knowledge corresponding to the set of search images;c) selecting a training image from t

[ What is claimed is:] [1.] A method of segmenting a set of images comprising the steps of:a) providing a set of search images, each search image containing a view of a target object;b) providing a priori domain knowledge corresponding to the set of search images;c) selecting a training image from the set of search images;d) obtaining at least one training contour from the selected training image, the training contour representing an outline of the target object in the selected training image;e) utilizing the selected training contour to extract training information corresponding to the target object in the selected training image;f) associating the training contour with a selected search image of the set of images, the training contour defining a search contour;g) deforming the search contour relative to the selected search image and maintaining predetermined information corresponding to the deformed search contour;h) utilizing the training information and the a priori domain knowledge to identify regions of the deformed search contour that correspond within predetermined limits to an outline of the target object in the selected search image; andi) repeating steps (c) through (h) for each image in the set of search images so as to segment all of the images in the set of search images.