强噪声下改进Canny算法的边缘检测

针对传统边缘检测算法抗噪性较差、易受噪声影响、误判率高和漏判等问题,提出一种强噪声环境下对传统Canny边缘检测算法的改进算法。该算法选用平滑聚类滤波取代高斯滤波对受噪声图像进行预处理;对滤波窗口内的像素点进行噪声检测,根据检测到的噪声点个数自适应调整滤波窗口的大小,改变窗口中各信息的输出,为图像中的重要信息赋予较大的权值,实现降低噪声影响的同时防止重要信息被过滤;极大值抑制阶段在3×3邻域内使用Sobel算子,额外加入45°、135°方向计算梯度幅值和方向,更全面地检测细节信息;针对图像的灰度变化使用平均方差来计算高阈值。仿真结果表明,在高斯噪声和椒盐噪声的混合强噪声干扰下,该算法得到的边缘提取结果明显优于传统算法得到的结果。

Edge Detection of an Improved Canny Algorithm under Strong Noise

In view of the problems of traditional edge detection algorithm,such as poor noise resistance,easy to be affected by noise,high rates of misjudgment and missing pixels,an improved algorithm for traditional Canny edge detection algorithm under strong noise is proposed.This algorithm uses smooth clustering filtering instead of Gaussian filtering to preprocess the noisy image.Noise detection is carried out for the pixel points in the filtering window.According to the number of detected noise points,the filter window can be adjusted adaptively,and the output of each information in the window can be changed to give a larger weight to important information in the image,so as to reduce the influence of noise and prevent important information from being filtered.The Sobel operator is used in the 3×3 neighborhood in the maximum suppression stage,and 45°and 135°directions are added to calculate the gradient amplitude and direction,so as to detect the detailed information more comprehensively.The average variance is used to calculate the high threshold for the grayscale changes of the image.The simulation shows that the edge extraction results obtained by the improved algorithm are better than those obtained by the traditional algorithm under the interference of Gauss noise and salt and pepper noise.