摘要： 根据彩色图像中像素点各颜色分量具有矢量类型数据的特点，提出用一种改进的开关型矢量中值滤波器(PFIVF)替换Canny边缘检测器中的高斯滤波器，以构建新的彩色图像边缘检测器。基于Peer Group技术和边缘检测方法建立PFIVF的多级噪声检测开关型结构，使PFIVF能够在滤除图像噪声的同时较好地保持图像的边缘细节。实验结果表明，改进的Canny边缘检测器能较好地抵抗高比例脉冲噪声。

关键词: Canny边缘检测器, 矢量中值滤波器, 噪声检测, 边缘类型, 高噪声比例

Abstract: By exploiting vector-class characteristics of color pixels, this paper proposes to use an improved Progressive Fast Impulsive Vector Filter(PFIVF) to replace the Gauss filter within the framework of Canny detector. The switching scheme of the PFIVF is constructed on the multi-level noise detection mechanism by using Peer Group technique and edge detection method, and makes it possible to preserve edge details of color image while suppressing noise. Experimental results show that the performance of the edge detector is robust to high-level impulsive noise.

Key words: Canny edge detector, Vector Median Filter(VMF), noise detection, edge type, high noise proportion

中图分类号: 

• TN911.73