关键词: 椒盐噪声, 决策分析, 非线性滤波, 轮胎痕迹

Abstract: Decision analysis can accurately distinguish between corrupted pixels and signal pixels. Mean filters can smooth the noise well, while adaptive median filters can preserve the details and edge information of the original image. In order to restore the tire trace image with high-density impulse noise, a new fast and efficient algorithm is proposed. This algorithm combines the merits of decision analysis, mean filter and adaptive median filter. It shows significantly better image quality than nonlinear filters such as traditional median filter and adaptive median filter. Experimental results show that this algorithm can eliminate high-density impulse noise in the gray tire trace image and color tire trace image at noise level less than 60%, and it effectively preserves the details and edge information of the original image.

Key words: impulse noise, decision analysis, nonlinear filtering, tire trace