一种图像中值滤波的加速算法研究

An Accelerated Algorithm of Image Median Filtering

针对传统的基于元素选择算法进行中值滤波速度较慢的缺点,考虑到中值滤波窗口中像素取值具有一定程度的重复性,在一趟处理过程中,增加了将处理区域内与主元值相同的元素移动到序列两端的处理,并判断主元或其重复值是否为中值:若是则可以提前确定滤波结果,若不是则可以将其全部排除,从而降低了下趟处理的数据量,提高了处理的速度。实验验证了该方法在不同噪声比例、不同滤波窗口及不同图像下的性能,结果表明该方法可以加快滤波速度,并且随着滤波区域的扩大性能更好。

Median filtering is one of the commonly used nonlinear filtering methods in image processing. Consid- ering the low speed of median filtering based on element selection algorithm and repeatability of pixel data of median filtering window, this paper moves the elements equal to the pivot to both ends of sequence and judges whether these elements or the repeated value are the median or not. If they are the median, the filtering result can be determined; if not, these elements can be excluded in next procedure. It can reduce the amount of data in the following processing steps and improve processing speed. Experiments verify the performance of this method with different noise ratio, different filtering window sizes and different images. The result shows that this method can accelerate the filtering speed and improve its performance with the expansion of filtering area.