针对彩色图像中脉冲噪声的滤波问题,分析了经典矢量中值滤波方法的原理和特点,提出了一种新的开关型矢量中值滤波方法。该方法通过中心像素的排序位置和矢量中值滤波与中心像素判的距离判断噪声存在的可能。实验证明,该方法在不同的噪声比例下均优于矢量中值滤波,比较其他开关型滤波方法具有很好的稳定性,能够更好地保持原图的细节。
The paper analyzed the principle and feature of the classical vector median filter, which was usually used to remove the impulsive noise from color images. It presented a new switching vector median filter using two threshold values.The new method determined the noise probability of the central pixel’s rank position in sliding window and the distance to the lowest ranked vector. The experiment shows that the filter has robust performance over the vector median filter and others at different noise level and can better keep fine details of original images.