一种新的彩色图像脉冲噪声滤波算法

摘要

In light of the practical requirement of embedded machine vision application system on filtering processing system,we proposed an improved vector median filtering algorithm for colour image.It uses comprehensively the grey correlation analysis method and vector median filtering technology.First it calculates the similarity degrees of each pixel vector in the image according to grey correlation analysis for measur-ing the correlation degree between the pixels,that is,to substitute the common space distance with grey correlations between pixels.Second-ly,it adds noise detection unit to the algorithm by using the grey correlation degree between pixels.By comparative experiment,it is found that the grey correlation analysis-based colour image filtering algorithm can quickly and efficiently filter the impulse noise in the image,and is superior to other algorithms in preserving effect of edge details in the image,and for computation efficiency,it is much higher than other algo-rithms as well.%针对嵌入式机器视觉应用系统对滤波处理算法的实际要求，提出一种改进的彩色图像矢量中值滤波算法。综合应用灰色关联分析方法和矢量中值滤波技术。首先根据灰色关联分析计算图像中各像素矢量之间相似程度来度量像素间的关联程度，即使用像素点间的灰色关联度代替常用的空间距离。其次，利用像素点间的灰色关联程度在算法中增加了噪声检测环节。通过进行对比实验，基于灰色关联分析的彩色图像滤波算法可以快速、有效地滤除图像中的脉冲噪声，对图像中边缘细节的保护效果要好于其他算法，而在计算效率方面更远高于其他算法。