一种Roberts自适应边缘检测方法

A Roberts' Adaptive Edge Detection Method

针对Roberts算法对噪声比较敏感且需要人为指定阈值等问题,提出了一种Roberts自适应边缘检测方法.利用Roberts算子的基本原理,扩充了检测方向,再根据待检像素周围的3×3像素邻域的平均像素灰度值,结合人眼视觉特征自适应地生成动态阈值,这样既保留了原Roberts算子的可并行处理、快速运算、边缘较细等优点,还对噪声有一定的抑制作用.同时,针对Roberts算法边缘检测相对粗糙、边缘细化效率较低的问题,分析和改进了原有边缘细化算法,即先对含有噪声的图像进行边缘检测,过滤其伪边缘,再对图像边缘进行细化处理,从而得到单像素边缘.实验比较表明,所提算法能够自适应地生成动态阈值,提高图像边缘细节信息的提取性能.

An adaptive edge detection method is proposed for the sensitivity to noise and given threshold of existing Roberts algorithms. The basic principle of the Roberts algorithm is used and the direction of the detection is expanded. The algorithm generates automatically adaptive threshold according to the mean value of the 3×3 area pixels around the detecting pixel and the property of people＇s vision. It not only can keep advantages of existing Roberts algorithm such as parallel process, fast calculation speed and the comparatively thin edge, but also can play a certain part in restraining noise. Because the edge detected by Roberts algorithm is comparatively thick, and the efficiency of the edge-thin algorithm is lower, the edge-thin algorithm is analyzed and improved. The improved algorithm detects the edge of the image where there exists noise, filters the fake edge, then thins the edge of the image, and finally gets the single pixel edge. By comparing the test results of the adaptive edge detection algorithm with those of the existing Roberts algorithm, the adaptive edge detection method can generate automatically adaptive threshold, improves the performance in getting detailed-edge and restraining noise.