Research On Image Denoising Algorithm In Machine Vision System

Machine vision technology is a comprehensive discipline, it is the latest development of the digital image processing and artificial intelligence. With the development of science and technology, the application of machine vision systems is becoming increasingly widespread, and the technology is becoming more and more mature. Image denoising is an important research direction in the field of machine vision. In this paper, image denoising in machine vision system is researched in depth.This document completed the overall design of presswork quality-defects system based on machine vision, including hardware design and software design.Then, the paper focuses on the image denoising algorithm. Summarizes the advantages and disadvantages of several classical denoising algorithms, and summarizes the image denoising evaluation methods of commonly used. Subsequently, the paper describes the non-local means denoising algorithm detailedly. The denoising effect of non-local means algorithm is very good. This method is very good at protecting the image edge information, but the computational complexity of this algorithm is high. We propose an improved method to reduce the computational complexity of this algorithm. The principle is: natural images usually contain a lot of gray value "flat" areas, the paper distinguishes these"flat"regions and elsewhere by gradient, that is, the image is divided into large gradient regions and small gradient regions. We can use local mean method to denoise in the small gradient region. In the large gradient regions, judge the similarity between two neighborhood by neighborhood average and the neighborhood average gradient angle, and then, remove the neighborhoods of low similarity.Through the experiment, verify the improved algorithm has better denoising effect from the subjective assessment and objective evaluation of image quality. The algorithm is better than the original non-local means denoising method both in denoising effect and the speed of execution. Finally, the algorithm used in the presswork quality-defects system, we will find the image quality is greatly improved after the processing by this algorithm.