基于视觉显著性和活动轮廓的图像分割

Image Segmentation Based on Visual Saliency and Active Contour Model

提出一种视觉显著性和传统的C-V模型相结合的图像分割方法,该方法首先提取图像的显著图,然后使用改进的自适应阈值法将显著图进行二值分割并提取边缘,并以此边缘作为C-V模型演化的初始轮廓.这样对于具有复杂背景的图像C-V模型可以从靠近目标物体的位置开始演化,从而得到较为准确的边缘,同时,也可以减少C-V模型的迭代次数.

An effective object segmentation method is proposed which combines C-V model with saliency map, It first extract image saliency map, then gets the general contours of the object using the improved adaptive threshold segmentation, makes this contours as the initial curves of C-V model. When the background of the image is complex this method ensures the active contours evolve close to the object to obtain more accurate edge and reduce the number of iterations of the C-V model. The experimental results show that the segmentation accuracy and efficiency of the algorithm are better than the C-V model both for the images which have the obvious object.