关键词: 图割, 图像分割, 特征向量, FCM聚类, 最大流最小割

Abstract:

A combination of multiple image feature information and graph cut algorithm is proposed for segmenting target by introducing multiple feature image information into the graph cut algorithm to reduce the negative influence on the target image of noise, complex background and other factors in the segmentation process. First, multiple image features are selected to compose the feature vectors, and the feature vectors of labeled target and background seed points are clustered by FCM. Second, the shortest distance of each pixel to the cluster center of each seed point of the two types is calculated, according to which the energy function is constructed. Finally, the maximum flow minimum cut method is used to get the results of image segmentation. Experimental results show that the proposed algorithm significantly improves the segmentation results over the traditional graph cut algorithm.

Key words: graph cut, image segmentation, feature vector, FCM clustering, max flow min cut