| 摘要 |
<p>The adaptive image segmentation system has a genetic learning system connected to an image analyser, and an image segmenter connected to the learning system. A segmented image evaluator is connected to the image segmenter and to the genetic learning system. The image analyser has an input for images and has an output for image statistics. The genetic learning system has one input for image statistics, a second input for external variables, an output for control parameters and a third input. The image segmenter has one input for images, a second input for control parameters and an output for segmented images. The segmented image evaluator has an input for segmented images, and an output for segmentation quality indications, connected to the third input of the genetic learning system. The external variables are environmental factors under which the images are acquired. The control parameters are segmentation parameters. The segmentation quality indications are used for initiating adaptation to incoming images which is based on differences in characteristics of previously analysed and of images currently being analysed. @(104pp)</p> |
