结合循环迭代方法的自适应中值去噪新模型

New adaptive median denoising model combined with cyclic iterative method

摘要:

针对传统去除椒盐噪声方法在图像噪声密度较高时去噪性能较差的缺点，本文提出了一种结合循环迭代方法的自适应中值去噪新模型，提高了高密度椒盐噪声下去噪算法的性能.该滤波器工作模式可分为三个阶段：首先，图像疑似噪声点预处理，通过极值判断法，将待处理像素点进行区分得到疑似噪声点；其次，确定噪声点处理，将已确定噪声点用邻域内的中值或均值自适应替换，从而完成去噪；最后，疑似噪声点再次处理，通过算法内置参数和条件，进一步判断疑似噪声点是否为噪声点.该模型还加入噪声标记点方法，通过迭代处理判断标记点结束去噪，得到滤波后的图像.仿真实验证明，本文提出的方法较传统的几种去除椒盐噪声滤波算法，针对无论是低密度噪声图像或是高密度噪声图像，去噪性能都有一定的提升，且能够较好地保留图像边缘和纹理等结构信息.

Abstract:

Aiming at the disadvantage of traditional salt-and-pepper noise removal methods that the denoising performance is poor when the image noise density is high, a new adaptive median denoising model combined with cyclic iterative method is proposed, in order to improve the performance of denoising algorithm in high-density salt and pepper removal. The working mode of the proposed filter can be divided into three stages. First, the image is preprocessed, that is, the suspected noise points are obtained from the pixels to be processed using the extreme value judgment method. Secondly, the noise points are determined and replaced adaptively by the median or mean value in the neighborhood to complete the denoising. Finally, the suspected noise points are processed again, and whether the suspected noise points are noise points is further judged by the algorithm with built-in parameters and conditions. The noise mark point method is also induced, and the filtered image is obtained by finding the end of mark point denoising through iterative processing. The results of simulation experiments show that the proposed method has a certain improvement in denoising performance for both low-density noise images and high-density noise images, compared with several traditional salt-and-pepper noise removal filtering algorithms.