Research On Some Key Issues In Image Denoising And Image Registration Algorithm

Image denosing and image matching are the two core subjects in the field of image processing.The current image denoising algorithm cannot remove noise well,especially when the noise is large.The natural images often contain weak edge and weak texture details which are the important information of the image.The existing image denosing algorithms can't keep edge and texture details well.And the traditional image matching methods still have some deficiencies such as the correct matching rate is low and the correct matching number is not enough.This paper is based on these problems,and research image denoising and image matching combined with fractional calculus.The main work is as follows:1.Non-local means(NLM)algorithm has good characteristic for removing noise and and preserving image details.But the algorithm is time consuming and the accuracy decreases significantly with the increase of noise.Fast Non-local means(FNLM)algorithm speeds up operation and reduces time cost,but the performance of denoising has not improved when noise increased.Aiming at the problem,this paper proposes a novel non-local means denoising method.We put forward a new exponential-Turky kernel function by combining Turky function and exponential function,and substitute the original exponential kernel function in both NLM algorithm and FNLM algorithm.Furthermore,both the Structure Similarity(SSIM)and Euclidean distance are introduced to measure the similarity between image neighborhood,which make the selection of weight more reasonable.The experimental results demonstrate that the proposed method improves denoising capacity greatly,especially for image with large noise.Additionally,the efficiency of proposed method is enhanced obviously against NLM algorithm,and the time complexity is equal to FNLM algorithm and time consumption is close to FNLM algorithm too.2.The Total Variation(TV)denoising model is unideal for maintaining the weak edge and weak textural detail of images,even though this model has satisfied the noise reduction performance.Adaptive Fractional Total Variation(AFTV)algorithm was presented to identify the texture and non-texture areas in an image based on local image information.The soft threshold value in an adaptive method was also calculated.Thus,the weak edges and weak texture details in the noise image can be preserved substantially better than the traditional TV algorithm could.However,the retaining effect of the weak edge and weak texture detail deteriorates as the noise increases,thereby substantially causing an evident staircase effect.Method To address this problem,the current study proposes a novel fractional TV denoising algorithm by applying fractional differential theory combined with the TV and characteristics of the residual image.Result A fractional-order TV model is proposed to substitute for the original first-order TV model.The image is divided into texture and flat areas based on the accurate local variance of the residual image,thereby enabling the adaptive selection of the fidelity item parameter to become considerably reasonable.Conclusion Experiments were performed on images with different noiselevels.In summary,the proposed method is capable of improving the denoising performance caused by the staircase effect for images with severe noise while preserving the weak edge and texture detail more effectively than the TV and AFTV denoising models could.3.In order to improve the correct matching rate and increase the number of correct matches,a new adaptive fractional-order sift matching algorithm is proposed.Firstly,according to the local information of the image,the tanh function is selected as the prototype,and an adaptive fractional order mathematical model is constructed.The optimal order of each pixel in the image can be adaptively calculated according to the local information features of the image,and a reasonable fractional-order mask can be constructed;replacing the original SIFT descriptor with a robust GLOH(Gradient Location and Orientation Histogram)descriptor;using the PSO(Position Scale Orientation)Euclidean distance instead of the traditional Euclidean distance as the measure of similarity of feature descriptors to effectively increase the number of matching points;and finally optimizing method with RANSAC Algorithm,eliminating the mismatch points,so that improving the matching accuracy.