| 摘要 |
<p>A similarity measurement method for the classification of medical images into predetermined categories. A small set of pre-classified images is required to employ the method. The images can be real world images acquired using a camera (32), computer tomography, etc., or schematic drawings as a source of images allows a quick test (34) of the method for a particular classification problem. The eigenvectors for each category are mathematically derived, and each image in each category is represented as a weighted linear combination of the eigenvectors. A test image is provided and projected onto the eigenvectors of each of the categories so as to reconstruct the test image with the eigenvectors (38).</p> |
