结合角点信息GVF Snake模型的图像分割方法

Image Segmentation Method Based on Corner Information and Adaptive GVF Snake Model

GVF Snake模型在边缘检测和图像分割中应用广泛.GVF Snake模型相比于传统的Snake模型有更大的捕捉区域和更好的分割性能.然而,GVF Snake模型在分割带有尖角的目标时曲线很难收敛到尖角处.针对GVF Snake模型的不足之处,本文提出角点信息和GVF Snake模型相结合的图像分割方法.首先,运用基于边缘轮廓曲率的角点检测方法,检测出图像中的真实角点,局部修正GVF场,然后结合角点信息给出局部角点力,最后将角点力与修正后的GVF场相结合的得出一种新的外力.实验证明,本文改进的GVF Snake模型能够更好的收敛到图像的尖角处.

GVF snake model is used widely in edge detection and image segmentation. GVF Snake model has larger capture range and better segmentation performance than traditional Snake model. However, the GVF method is inaccurate in capturing the sharp comer of the object during image segmentation. To solve the inadequate of GVF Snake, a new technique of GVF Snake model based on angular point information to extract boundaries of objects having sharp comers is presented. First, we check the real comer points at the edge u- sing curvature-based comer detector, and locally modified GVF force field. Then local angular point force is put forwarded based on comer information. Finally,combining the comer force and the modified external force of the GVF Snake,a new external force of improved GVF Snake is proposed. Experiments indicate that the new algorithm have a better convergence to the sharp comers.