基于细节保留的椒盐噪声自适应滤波算法

Salt & Pepper Noise Adaptive Filtering Algorithm Based On Detail Preserving

针对灰度图像中椒盐噪声的特点,提出了一种更加精确的噪声检测方法:该方法利用滤波窗口内像素点灰度值的不同,将受椒盐噪声污染的图像中像素点划分为噪声点,疑似噪声点和信号点。通过设定阈值,并参考相邻像素点的相关性来进一步区分疑似噪声点,最终建立噪声标记矩阵。对于被标记的噪声点,采用自适应滤波算法,保留更多的图像细节。仿真结果表明,该算法在除去噪声点的同时,对于边缘细节也有非常好的保护作用。

According to the characteristics of salt and pepper noise in gray-scale image, a more accurate noise detection method is proposed. From the difference of pixel gray value in filtering window, each pixel is classified to be signal pixel, noise pixel and possible noise. By setting the threshed value and considering the correlativity of adjacent pixels, possible noise pixels are subdivided into noise pixels and signal pixels, and the noise matrix is eventually established. For the marked noise points, using adaptive filtering algorithm, the restored image can keep more details. Simulation results show that the results of salt and pepper noise filtering are improved and the detail-preserving are also greatly improved.