| In past almost twenty years, active contour model, as an important research field in computer vision, attracts the attention of more and more researchers. It is a top-down processing with prior knowledge, which is different from the classical computational vision proposed by D. Marr. It provides a theoretically uniform framework to a series of problems, such as contour extraction, stereo matching and object tracking. And it also has been successfully applied to image segmentation, medical image processing, human-computer interaction and many other research and practical fields. But it is still necessary to study this model thorough and propose new model or algorithm, for some inherent disadvantage.This paper illuminates the foundation of active contour model, and studies the improved algorithm of energy minimizing of traditional parametric active contour model. If original contour is not placed close to the real boundary of object, the result would be greatly affected by some effects such as image noise and strong edge, and would be inaccurate. To solve the problem, firstly, we assume different Gaussian distribution for object and background. Then, a new region energy based on Bayes theory is introduced into the traditional parametrical active contour model. To decrease the effects from strong edge, regional statistical information and an adaptive strategy are introduced into the tension energy. In the last, the modified model combined by motion estimation on the basis of block-matching method, is used in object segmentation with great noise and rigid object tracking in clutter.The experimental results show that the extent of contour affected by noise is reduced and the accuracy of segmentation is improved to some extent after the new region energy is introduced. At the same time, the region information and the adaptive strategy makes the effects of tension force opposite in the region of object and background. This overcomes the shortcoming of region energy to a... |
PDF Full Text Request