Abstract: To remove the random impulse niose introduced during the signal trsansmission and acquisition, a new vector median filter algorithm based on fuzzy reasoning is proposed. The approach utilizes the fuzzy rules to judge whether details exist in current filtering window. Different filter methods are adopted according to the fuzzy reasoning result. When not being regarded as noise pixel, the original pixel is retained. The experiments show that the fuzzy filter algorithm outperforms other state-of-the-art filters.

Key words: vector median filter, random impulsive noise, color image

摘要： 为有效消除彩色图像在采集和传输过程中出现的随机脉冲噪音，提出了一种基于模糊推理的矢量中值滤波算法。该算法通过设计的模糊规则判断当前滤波窗口对应是否为细节区域，采用不同参数的经典矢量中值滤波进行处理。若未被判断为噪音时，原滤波窗口中心像素会保留。通过与目前常见的各类矢量滤波算法进行比较，该算法在图像细节的保留上有较好效果。

关键词: 矢量滤波, 随机脉冲噪音, 彩色图像