Research And Application Of Image Denoising Algorithm Based On Salt And Pepper Noise

Due to its advantages of high efficiency,high sensitivity,and non-contact measurement,surface roughness detection based on machine vision has been rapidly developed in the detection of surface roughness of work pieces in recent years.In the process of machine vision image acquisition,transmission and storage,due to various internal and external reasons,the image will be subject to various noise interference,resulting in decreased image quality,among which salt and pepper noise is a relatively common type of noise.Therefore,how to filter out salt and pepper noise and improve image quality is one of the key issues in image processing and machine vision inspection.The traditional adaptive median filtering algorithm has achieved a certain effect in filtering salt and pepper noise,but it is easy to misjudge the high frequency signal point as noise point,single de-noising window direction,and middle value replacing noise point caused by blurred image boundary and other problems.The dissertation proposes an improved adaptive median filtering algorithm aiming at the deficiency of tra ditional filtering algorithm for salt and pepper de-noising.The dissertation further studies the high-density salt and pepper noise filtering method,focuses on the noise point detection method,and proposes a high-density salt and pepper noise filtering algorithm.The paper applied the subjective evaluation method and the objective evaluation method to verify the proposed algorithm.In this thesis,the improved adaptive median filtering algorithm will be applied to the detection of surface roughness of ge ar tooth surface roughness,and the effectiveness of the method was verified.The main research work of this paper is as follows:(1)For the traditional adaptive median filtering algorithm,it is easy to misjudge high-frequency signal points as noise points.This paper applies two-level threshold method to noise detection.For the single direction of the filtering window,this paper introduces a full-scaled and sub-window for filtering.For the purpose of blurring the boundary of the image due to the media n replacement of noise points,the minimum variance theory is applied to the filtering window.The variance of different subwindows is calculated first,and then the noise point is replaced by the median of the minimum variance subwindows.Through experimental simulation,this paper finally validates the effectiveness of the improved algorithm.(2)For the high density of salt and pepper noise,some classic filtering algorithms and improved adaptive median filtering algorithms have reduced filtering effects.To solve this problem,this paper focuses on the noise detection method,introduces a noise marker matrix,and proposes a high-density salt and pepper noise filtering algorithm.Firstly,the algorithm uses the noise detection method to divide the image pixels into signal points,suspected positive noise points and suspected negative noise points,and marks them with 0,1,and-1,respectively,to construct a noise marker matrix.Next,by calculating the local statistical characteristics of the noise marker matrix,the suspected noise points are specifically divided into signal points,noise points and indefinite points.Finally,different countermeasures are taken for filtering.Simulation analysis verifies the effectiveness of the method.(3)The machine vision based on surface roughness detection method is mainly subjected to low-density pepper-salt noise during image acquisition on the gear surface.In this paper,the improved adaptive median filtering method proposed in this paper is applied to the preprocessing of gear surface roughness for the calculation of gear surface roughness and evaluation indicators.By comparing the original image and the denoised image,this paper verifies the effectiveness of the method.