典型微分算子的图像边缘检测对比研究

A comparative study of image edge detection with typical differential operators

基于图像梯度灰度值的微分边缘检测是一种常用的边缘检测方法,其原理是根据图像像素点梯度变化确定图像中的边缘信息,将图像中的边缘轮廓检测出来,可以在保留图像重要信息的同时减少数据量.但不同的微分边缘检测算子在检测的准确性、连续性、边缘宽度以及抑制噪声等方面存在较为明显的差异性.因此,采用典型微分边缘检测方法中的Roberts算子、Prewitt算子、Sobel算子、Laplace算子和Canny算子分别对原始图像和添加噪声并滤波的图像进行边缘检测,其中噪声是方差为0.01的高斯噪声,检测结果显示Canny算子的边缘检测效果较好,且其峰值信噪比为15.035,均方误差为0.038,信噪比为2.837,表明该微分算子抗噪声干扰能力最强,但缺点是所用时间较长,约为0.57 s,是其他算子的5倍以上.

Differential edge detection based on image gradient gray value is a common edge detection method.Its principle is to determine the edge information in the image according to the gradient change of image pixels,and detect the edge contour in the image,which can reduce the amount of data while preserving the important information of the image.However,different differential edge detection operators have obvious differences in detection accuracy,continuity,edge width and noise suppression.Therefore,Roberts operator,Prewitt operator,Sobel operator,Laplace operator and Canny operator in typical differential edge detection methods are used to detect the edge of the original image and the image with added noise and filtering respectively.The noise is gaussian noise with variance of0.01.The detection results show that the edge detection effect of Canny operator is good,and its peak SNR is 15.035,mean square error is 0.038,SNR is 2.837,indicating that the differential operator has the strongest ability to resist noise interference,but the disadvantage is that the elapsed time is about 0.57 s is five times more than any other operator.