Abstract: This paper proposes an edge detection algorithm based on gray level difference analysis. The processing of the novel edge detector not only is isotropic but also educes ‘gray-level ridge’ to insure the results to be continuous and simultaneously reduces the detector’s sensibility to noise. Optimal non-maximum condensation method is used to effectively guarantee only one response to a single edge. Experimental results prove that the proposed approach are given works well over most images, and is fast enough for real time applications

Key words: Image processing; Edge detection; Pattern recognition; Gray level difference

摘要： 提出了一种新的基于灰度差统计的边缘检测方法。该方法基于全图的灰度差统计获得初步边缘判定阈值，可以降低边缘检测结果对噪声的敏感程度，并且不需要进行边缘方向估计，而是引入灰度脊跟踪优化算法，使边缘检测结果具有连续性和准确性。由于使用了检测结果优化算法，有效地避免了一条真实边缘对应多条边缘检测结果的情况。实验证明该方法应用在大部分真实图像的边缘检测效果优异，同时运行速度能满足大部分边缘检测应用的需要。

关键词: 图像处理；边缘检测；模式识别；灰度差