Median filtering method and device

Abstract

The invention provides a median filtering method and device. The method comprises the steps that the first kind pixel value similarity and the distance attenuation factor value of neighborhood pixelsand a center pixel in a first image are calculated; according to the first kind pixel value similarity and the distance attenuation factor value, the median value of the first kind weights of the neighborhood pixels is calculated; the median value, which is corresponding to the median value of the first kind weights, of second kind pixel values is found; and weighted summation is carried out on the second pixel values of partial pixels in the second kind pixel values of neighborhood pixels in a second image to acquire a weighted summation value, wherein the second pixel values of partial pixels are close to the median value of the second kind pixel values. According to the median filtering method and device, the first image is combined to perform weighted median filtering on the second image; the method can better maintain and optimize image edges without edge blurring and noise diffusion compared with a conventional multi-image filtering method; and the denoising ability and the edgeretention ability of the method are superior to the denoising ability and the edge retention ability of a traditional single-image filtering method.

Description

technical field [0001] The invention relates to stereo vision processing technology, in particular to a median filtering method and device. Background technique [0002] Image filtering methods can be divided into two categories according to the processed objects: one is single-image filtering methods, such as mean filtering, median filtering, bilateral filtering, etc.; the other is multi-image filtering methods, such as joint bilateral filtering and guided filtering. filtering etc. The processing object of the single-image filtering method is a single image, and the auxiliary information of other images such as guide images cannot be used. Therefore, when there are guide images, its filtering ability is limited. Since the multi-image filtering method uses the information of the guide image, its denoising ability and edge preservation ability are generally better than the single-image filtering method. [0003] Most of the filters used in traditional multi-image filtering ...

AI Technical Summary

Problems solved by technology

[0003] Most of the filters used in traditional multi-image filtering methods such as joint bilateral filtering and guided filtering have smooth characteristics, which are easy to lose image details and diffuse image edges and noise.

Images