Noise-detection-based high density impulse noise self-adaptive filtering algorithm

Abstract

The invention relates to a noise-detection-based high density impulse noise self-adaptive filtering algorithm (PA). For the instance of pepper salt impulse noise, the thought that detection is performed before filtering is adopted, first, noise points and signals points are discriminated according to pepper salt noise images, and then the noise points are subjected to self-adaptive recovery according to a certain criterion. The method includes the specific steps that first, nine pixels in a W3(i, j) window are sorted in an ascending mode according to a gray value; second, the point (i, j) with the corresponding F(i, j) being 0 in a noise image X is subjected to self-adaptive filtering; third, the next (i, j) is converted into the first step to be processed from top to bottom and from left to right till the last pixel point is filtered, wherein a critical point is not processed. Through a large number of simulation tests and quantitative evaluation index comparison, the algorithm has the advantages that detailed information of the image is kept well while high density image noise filtering is performed and time cost is low compared with the same kind of algorithms.

Description

technical field [0001] The invention relates to a high-density impulse noise self-adaptive filtering algorithm based on noise detection, belonging to the technical field of noise detection. Background technique [0002] Noise filtering is the most important link in many image processing systems. Impulse noise is the most typical noise type, among which salt and pepper noise (bipolar noise) is the most common. Salt and pepper noise is black and white bright and dark point noise introduced by image sensors, transmission decoding and other links. For a 256-level grayscale image polluted by salt-and-pepper pulse noise, the salt-and-pepper noise corresponds to pixels with a maximum grayscale value of 255 (white salt grains) and a minimum grayscale value of 0 (black peppercorns). [0003] The standard median filtering (Standard Median Filtering, SMF) algorithm is to perform a filtering operation on each pixel, and replace the gray value of the filter point with the gray value of ...

AI Technical Summary

Problems solved by technology

In addition, the algorithm proposed above guarantees a low missed detection rate and false detection rate in the detection of suspected noise points, but due to the 2-level adaptive filtering calculation, the time cost is very high

At present, the noise filtering algorithm based on noise detection or the detection method of suspected noise points is complex, or the ability to filter high-density noise images is limited, which greatly limits the adaptability and real-time performance of this type of algorithm.

Images