图像平滑处理中,如何在去除噪声的同时完整地保留图像边缘细节一直是非线性滤波算法研究的热点问题.提出了一种基于极值的自适应中值滤波算法,该算法根据图像中某点是否为邻域极值点将全部像素分为可疑噪声与信号两类.对可疑噪声点采用包括八个一维窗口和一个二维窗口在内的不同尺度和不同方向的九个子窗口,按照各个子窗口的均方差大小,自适应选择窗口进行中值滤波;对信号点不加处理,灰度值不变.测试结果表明,该算法的滤噪特性和细节保护能力优于多级中值滤波;执行速度较快,优于经典中值滤??

In the smoothing process of image,it is very important for nonlinear filter to reserve fringe detail of image while filtering noise.A new adaptive median filter algorithm based on extreme value is presented in the paper. According to whether gray value of arbitrary pixel is its neighborhood extremum, all pixels are divided into doubtful noise pixels and signal pixels. The gray value of signal point is unchanged. Following the average variance of the nine multiscale and multidirectional windows including eight one dimensional windows and one two-dimension window, the gray value of noise point is selected adaptively. Test result indicates that new algorithm is superior to multistage median filter in performance of filter noise and detail preservation. At the same time, operating speed of the algorithm is faster than classical median filter.