其基本思想是：首先，用微分算子检测图象的边缘；然后，在这些边缘象素点上进行二值化阈值的自动选择；最后，对于其它非边缘象素点则采取常规方法进行二值化处理。实验结果表明，这个基于边缘特征检测算子的算法能很好地保留原图的边缘特征，并能处理低质量的图象-The basic idea is: First of all, the differential operator with the edge detection image　And, at these points on the edge pixel values for two of the automatic threshold selection　Finally, with regard to other non-edge pixel point to take conventional methods treatment of two values. The experimental results show that the edge-based feature detection algorithm operator can retain the good image of the edge feature, and can handle low-quality images