一种自适应的运动目标模糊边缘检测方法

提出了一种有选择的局部背景快速分割方法；针对模糊边缘难以检测的问题,提出一种将梯度法与灰度阈值法相结合的方法.该方法首先通过大津法确定目标的位置；然后在其邻域内求出关于灰度值分布的直方图的极值点,再通过灰度阈值法得到目标的初步边缘；最后根据梯度信息对边缘进行微调,减少了灰度不均匀等因素造成的影响.实验结果表明,文中的方法可以有效地提高模糊边缘的检测效率.

An Adaptive Method for Detecting Fuzzy Edges of Moving Targets

This paper proposes a fast and selective method for segmenting background of images.At the same time,an adaptive method,combing gradient method with gray scale threshold method is also proposed for the difficulties about detecting fuzzy edges.The new method firstly determine the locations of the targets with the help of the Otsu method and then find the extreme points of the histogram about the distribution of the gray value within these targets’ neighborhood.Gray scale threshold method can help to get the initial edges here and finally we fine-tune these edges according to their gradient information to reduce distraction caused by the non-uniform gray-scale and some other factors.Results of the experiment show that this method can effectively improve the efficiency of the detection of the fuzzy edges.