中文摘要：

针对痕迹图像在比对与识别中的预处理要求，以图像像素灰度和邻域信息的二维矢量为聚类样本，以二维直方图确定阈值数，采用对样本进行加权的模糊C均值聚类算法对痕迹图像进行多阈值分割处理．实验结果表明，基于样本加权的模糊聚类算法速度快，对痕迹图像有较好的分割效果．

英文摘要：

Aiming at the pre-processing requirement of mark image in comparing and distinguishing design, the pro- posed paper describes a two-dimensional vector with pixel grayscale of image and neighborhood information as cluster sample. The threshold is dependent on two-dimensional histogram. We give a multi-threshold segmentation processing / method to mark image using weighted fuzzy c-means clustering algorithm. The results of the experiment show that the weighted fuzzy clustering algorithm based on sample is rapid, and has better segmentation effect on mark image.