新的梯度边缘检测方法

Novel edge detection method based on gradient

提出了一种新的基于梯度算子法检测灰度图像边缘的方法。该方法在3×3的邻域窗内,分别用梯度法对窗内的水平方向、垂直方向和正、负45°角方向上的边缘线进行检测,找到梯度最大的边缘线,从而组成最终的边缘图像。实验表明:该方法简单,边缘检测图像效果好。在此基础上,又提出了增强边缘检测方法和简化边缘检测方法。这两种对基于梯度算子法的改进方法分别适用于检测小的边缘梯度和需要快速检测边缘这两种情况。

A novel edge detection algorithm based on gradient of gray intensity is proposed. In a 3×3 neighborhood window, the pattern edges in horizontal, vertical, 45°, and 135° are detected by gradient operator respectively, and then the biggest gradient of these four edges is regarded as final edge in this window. Experiments show that this algorithm is simple and efficient. Based on it, an enhancement method and a simplified method are proposed, which can detect more detailed edges and consume less computing time, respectively.