摘要： 针对标准中值滤波算法去噪能力不强的问题，提出一种基于噪声检测的图像去噪算法。通过原图像与4个方向核的卷积，将像素点分为噪声点和信号点，直接输出信号点，而只处理噪声点。若信号点所在的最小邻域中存在信号点，则利用该点进行中值滤波；否则，继续扩大邻域范围。实验结果表明，该算法不仅能有效去除图像噪声，而且能较好地保护边缘，提高恢复图像的清晰度。

关键词: 噪声检测, 方向核, 卷积, 中值滤波, 邻域, 噪声点

Abstract: The denosing performance of standard median filtering algorithm is poor. Aiming at this problem, this paper proposes a denoising algorithm based on noise detection. It uses four directions of the nuclear convolution to distinguish noise point and signal points from image pixels, diretly outputs the signal points, and processes the noise point. If the minimum neighborhood of the noise point has signal points, the signal points is uesed to median flitering; Otherwise, it expands the range of the neighborhood in value processing. Experimental results show that the algorithm not only can effectively removes image noise, but also can better protect the edge, which enhances the recovery image clarity.

Key words: noise detection, direction core, convolution, median filtering, neighborhood, noise point

中图分类号: 

• TN911.73