AN IMPROVED IMAGE DENOISING ALGORITHM BASED ON GREY RELATIONAL ANALYSIS
The problem of image denoising is a primary operation in the stage of image pre-processing. In recent years, combining the image filter with grey system theory, some scholars present an adaptive added-weigh mean filter algorithm which is a optimizing process to obtain the value of weigh decided by comparing the similarity between the grey levels of sample pixels in the filter window and the mean grey levels of pixels in the window based on the grey relational degree. Compared with the traditional mean filter, it has achieved certain progress. However, because of the defects of the algorithm of its own, when the noise density reaches a certain level, the processing results are unsatisfactory. The main part unreasonable is that when filtering the image we do not make a clear distinction between the noise points and normal ones, which will inevitably lead to the edge blurred phenomena. In this paper, we first use a new detector to distinguish between the noise and normal points. Then we just do filter operation to the pixels which have been sentenced to noise points using an improved filter algorithm based on grey relational analysis. Finally, the experimental results demonstrate that the algorithm improved is significantly better than the classical mean filter, median filter and the traditional image filter algorithm based on the grey relational analysis.
Image Denoising Grey Relational Analysis Noise Detect Neighborhood Window Filter
LI Gang XIAO Xin-ping CHEN Rui
School of Science, Wuhan University of Technology Wuhan, Hubei 430063, China School of Science, Hube School of Science, Wuhan University of Technology Wuhan, Hubei 430063, China School of Science, Hubei University of Technology Wuhan, Hubei 430068, China
国际会议
武汉
英文
480-484
2009-10-16（万方平台首次上网日期，不代表论文的发表时间）