边缘检测中的改进型均值滤波算法

Improved Mean Filtering Algorithm in Edge Detection

针对传统高斯滤波器算法存在边缘模糊及局部伪影的问题,提出一种改进型均值滤波算法,以有效去除图像中的高密度脉冲噪声。通过求解图像直方图,检测图像直方图的左右2个峰值点,以查找出可疑的噪声点,进而使用灰度值确定其坐标。利用图像像素点周围灰度值的相似性,在局部窗口选择递归加权平均值代替被污染的图像像素点灰度值。结合自动选择Canny算子的双阈值方法,对灰度图像边缘进行检测。实验结果表明,该算法的算子滤波效果优于LOG算子和传统灰度Canny算子,且具有较好的稳定性。

The traditional Gauss filter algorithm has the problems of edge blur and local artifact. So an improved mean filtering algorithm is proposed to effectively remove the high density impulse noise of the image. The left and right two peaks of the image histogram are detected by solving the image histogram, so as to identify suspicious noises and determine those coordinates with gray values. Recursive weighted average values in the local window are used to replace pixel gray values of contaminated images with the similarity of the gray values around the image pixels. Combined with the double threshold method which automatically selects the Canny operator, the gray image edge is detected. Experimental result shows that the operator filtering effect of the proposed algorithm is better than LOG operator and gray Canny operator, and it also has good stability.