基于中值的图像椒盐噪声的非迭代滤除

去除椒盐脉冲噪声是图像处理中的一个重要问题。标准的中值滤波能消除大部分椒盐噪声,但是却毁坏了图像中的部分细节,基于最大最小值的波峰波谷滤波虽在一定程度上对这个问题有所改善,但效果却不是很好。改进方法与基于最大最小值的波峰波谷滤波方法的区别是在如何探索噪声点上又考虑了中心象素与其邻域象素的相似性。实验结果证明了新方法在图像的峰值信噪比和噪声对原图像的毁坏程度两方面的优越性。

Effective and Non-iterative Removal Based on the Median Filter for Salt-Pepper Noise from Images

Salt-pepper impulsive noise removal is an important problem in image processing.The common median filter can get off majority of noise and thus destroys fine details.To some extent a variation of the peak-valley filter based on a recursive minimum-maximum method addressed this drawback,but the result is not enough.In this work,we consider the similarities between the center pixel and some of its neighbor pixels in search of corrupted pixels based on the above method.The experiment demonstrated the effectivity of the new method in both PSNR and destroyness of noise to original image.