Research On Methods For Removing Impulse Noise And Gamma Noise In Images

Digital image has become an intuitive and effective expression way for mankind to obtain information,however in the process of collection,storage,transmission and processing,images are inevitable to be degraded by various types of noises,which to generate unnecessary obstacles for the subsequent other relevant image processing work.Therefore,designing reasonable effective denoising method to filter out the noise in image has become a research hotspot in the field of image processing.There are many reasons for producing noises in images,according to the different imaging environments can have a variety of different kinds of noisy types,the impulse noise and gamma noise are two different types of noises produced by different imaging environments.In this paper,we consider these two kinds of noises as the main research objects,further to study the image denoising methods,to realize images pretreatment.Impulse noise is one of the most common noisy types in digital image,mainly caused by mechanical vibration or the interference of external environment results in electronic components arise sudden impulse signal,it only affects a part of pixels in image,while gamma noise is different from common impulse noise,it is mainly caused by scattering gamma photons in the process of neutron radiography imaging,these gamma photons reach to CCD chip and deposit energy to form white spots which differ in size,shape,intensity.Two kinds of noises both have the strong randomness,but as a result of mechanism of production and pixels distribution of noises are different,at present,applying the denoising method of impulse noise to filter gamma noise can only has inhibition effect and can't remove entirely,so this paper proposes two different denoising methods combined with the features of both noises.At first,for solving the ill-posed inverse problem in the removal of impulse noise,in this paper,based on image sparsity the regularization method is introduced,combined Tikhonov variational model with ???₀ norm minimization to get a new Tikhonov-???₀ denoising method for impulse noise,with the aid of alternating direction method of multipliers?ADMM?,the global minimum value problem of denoising model is decomposed into two sub-problems of local variables minimum values,and through repeated iteration,to obtain the optimal solution of denoising model.This method can remove high intensity impulse noise and also preserve more details,but it can't remove gamma noise effectively.Hence,to remove gamma noise in image,this paper introduces a new "detection-corrosion-recovery" repaired strategy,puts forward a recursive algorithm based on spatially adaptive filter.Firstly it detects noisy area in image,and then gets restored image through iteration corrosion method,this method overcomes the denoising method's shortage of impulse noise in gamma noise removal.The experimental results show that the above two different denoising methods based on different noisy types both have satisfactory denoising effect,no matter visual perception or PSNR values,not only in the denoising performance have good robustness,and in details preserving also have superiority.