图象分割是图象处理中的一项重要工作，目前手工与自动相结合的分割方法在实际工作中得到了广泛应用。本文根据图象经Maar变换后的特征，采用新的判断准则，提出了一种复杂物体边缘定位算法，可以对具有尖角特征的物体轮廓进行快速准确地提取，同时利用矢量化方法消除毛刺，使跟踪获得的边界更符合物体的实际轮廓特征。对多种图象的实验表明本文方法十分有效。-Image segmentation is an important image processing, the current combination of manual and automatic segmentation method in practical work has been widely used. Based on the image transformed by Maar features a new criterion is proposed positioning algorithm is the edge of a complex object, you can have the angular features of the contour to quickly and accurately extract, while taking advantage of vector-based approach removes the burr, to trace the border to obtain more in line with the actual contours of the object features. Images of a variety of experiments show that this method is very effective.