一种改进的基于梯度的自适应边缘检测算法

对常见的边缘检测算法进行改进,从梯度的定义出发,提出了一种应用于灰度图像的自适应阈值边缘检测算法.根据边缘处像素灰度值的差异,将模版中的9个像素分两组进行分析,计算出像素的梯度幅度和梯度方向.按梯度值的不同将图像分割成若干个区域,计算每个区域的灰度平均值,确定阈值,实现边缘检测.实验表明,该方法检测出的边缘更细、更准确,可以除去虚假的边缘,是一种有效的对灰度图像进行边缘检测的方法.

An Improved Self-adaptive Algorithm of Edge Detection Based on Gradient

This paper improves on the common edge detection algorithm, and presents an edge detection algorithm of self-adaptive threshold applying to gray level images based on the definition of gradient. According to the different gray level of the pixels on the edge, the nine pixels in the mask were divided into two groups for analysis, the gradient magnitude and gradient direction of each pixel was calculated. The image is divided into several regions by the different gradient values. The edge detection can be implemented when we figure out the average gray level for each region and determine the threshold. The experimental result shows that this method can detect thinner and more accurate edge, it can also eliminate the false edges. It is an effective method to detect the edge for the gray level image.