运用模糊连接理论分割非连通图象

Unconnectedness Image Segmentation Based on Fuzzy Connectedness Theory

本文提出了一种利用模糊连接理论讨论非连通图象的分割方案．该方案较好地利用了图象本身的模糊特性，修正了模糊连接强度的定义，从而成功地实现了分割数字、字母和汉字这样一些连通的或非连通的图象．它克服了Ｊ．Ｋ．Ｕｄｕｐａ算法仅能分割连通图的缺点．同时论文中还讨论了分割灰度图象和彩色图象的模糊连接强度定义及有关参数的选择方法．实验例子证明了这种方法的有效性．

This paper proposes a method of image segmentation based on Fuzzy connectedness theory. This method uses fuzzy features of the pictures combined with threshold comparison. We also demonstrate the realization of segmenting mixture numbers and characters and Chinese words which is linked up and not linked up. It overcomes the faults of only segmenting the linked up pictures by J.K.Udupa's algorithm. We also give the definition of fuzzy connectivity intensity of color image and the parameters using for segmenting images. Experiment result shows the effectiveness of our method.