基于先验知识改进Snake模型的脸部特征分割

针对传统Snake模型在脸部特征分割中遇到的边界泄漏问题，给出两种局部信息用于边缘增强，改进了外力场；引入了预测／校正两步形变Snake模型的概念，将Snake，模型在内力和外力作用下的形变结果作为Snake轮廓新位置的预测，然后用目标形状的先验知识对预测结果进行校正，使得Snake的形变由预测／校正两步来完成。在分割脸部特征时，构造了对Snake形变进行校正的准则。实验结果表明，该预测／校正两步形变的Snake模型对脸部特征分割有较好的效果。

Facial Feature Segmentation via Snake Model Improved with Prior Knowledge

Concerning the boundary leaking problem of traditional Snake model in facial feature segmentation, a new two step deformation Snake model with additional two kinds of local information to enhance the weak boundary and applying a new external force field is proposed The Snake deformation results under internal and external force are taken to predict the new contour position and the prior shape constraints of object are used to correct the predicted results, such that the Snake deforms in a two step mode of prediction correction The correction rule is constructed specially for facial feature segmentation Experimental results validated the performance of this two step approach for facial feature segmentation