Research On Optical Coherence Tomography Image Denoising Algorithm Based On Singular Value Shrinkage

Optical coherence tomography(OCT)is a micron-scale,non-invasive,non-invasive,high-resolution optical imaging mode.It has a wide range of applications in the fields of ophthalmology,dermatology,oncology,dentistry and cardiovascular medicine and other medical disease detection fields,as well as cultural relic restoration,jewelry detection,and industrial detection.However,due to the low coherent interference of light,the OCT image is in-evitably damaged by speckle noise,which leads to a significant reduction in the quality of the OCT image.For example,the structure and details of the image are blurred,and the sharpness and contrast of the image are reduced.Low-quality OCT images have adverse effects on subsequent image process-ing,such as increasing the difficulty of doctors in detecting and analyzing lesions,and increasing the error rate of edge detection and segmentation.Therefore,it is of great value and significance to remove speckle noise from OCT images and improve the sharpness of OCT images.In order to remove the noise on the OCT image,researchers have pro-posed various methods to deal with it.These methods are mainly divided into two main categories:one is hardware-based modification methods;the other is software-based digital filtering methods.The hardware-based modification method is to improve the existing OCT system and related system hardware and incident light sources in the system,which can reduce or eliminate part of the noise in the image before the final OCT image is obtained.However,this method of improving the hardware is too complicated and expensive,and it is difficult to apply it to practical commercial OCT systems.The software-based digital filtering methods are more economical and effective,such as methods based on total variation and diffusion equations,wavelet decomposition transforms,probability statistics,non-local self-similarity,s-parse representation,low-rank and network.However,these algorithms do not fully consider the relevant characteristics of OCT images,so it is difficult to remove the speckle noise of OCT images while preserving its structure and details.Considering the relevant information of OCT,this paper mainly studies several digital filtering algorithms with low-rank approximation.(1)This paper proposes an adaptive singular value shrinking speckle attenuation algorithm based on generalized likelihood matching for OCT imagesAiming at highly sparse redundancy of OCT image data intensity,this algorithm uses the generalized likelihood ratio as the similarity criterion for block matching to construct a low-rank group matrix,and then uses the sin-gular value decomposition and shrinkage strategies to reconstruct the OCT estimated image.Finally,feature adaptive backward projection strategy is adopted to better recover the underlying layered structure and detail infor-mation of the OCT image.Compared with several existing advanced algo-rithms,the experimental results demonstrate that the proposed algorithm achieves a state-of-the-art despeckling performance in terms of both quanti-tative measurement and visual interpretation(2)This paper proposes a feature-oriented two-stage singular value shrink-ing denoising algorithm for OCT images based on a low-rank approximation framework.First,a weighted absolute distance is employed to find nonlocal similar patches that exhibit high correlation to a given reference one.Next,the singular values of the group matrix formed by similar patches are shrunk by mixed thresholding so that they are closer to the singular values of that of la-tent noise-free image.Finally,an iterative regularization technique is adopt-ed to improve the denoising performance of the proposed method,where the backward projection parameter in each pixel is adaptively determined by its corresponding gradient variation.By using these strategies,the proposed method not only effectively removes the speckle noise of OCT images,but also preserves fine structural information of objects.Experimental results show that the proposed algorithm is competitive with some state-of-the-art speckle removal techniques in terms of both objective metrics and subjective visual inspection.(3)This paper proposes a speckle reduction algorithm for OCT images based on fractional filtering and enhanced singular value shrinking.An OCT image is first divided into many overlapping image blocks and each block is filtered using a fractional mask,and then an absolute distance is used as a similarity criterion for block matching to form a low-rank group matrix.Furthermore,the fractional-order preprocessing is performed on the group matrix.Finally,singular value decomposition,a piecewise Laplace shrinkage,aggregating and boosted iterative regularization technique are used to reconstruct a filtered image.Extensive experiments are performed on 18 OCT images of the retina of the human eye to verify the validity of the proposed method.Experimental results show that the proposed method harvests best PSNR,SSIM and EP results in most cases.In addition,results of the paired samples t-test show that the proposed method can remove noise more thoroughly and better preserve the structural information of the OCT images.(4)This paper proposes an iterative denoising algorithm for OCT images based on chi-square similarity and fuzzy logic.An OCT image is first divided into a lot of overlapping image blocks,and a Chi-square distance similar block matching is utilized to form a low-rank group matrix.Then,the singular value decomposition of the group matrix is performed,and the singular values are contracted by different weights with fuzzy logic.Finally,a pixel intensity fuzzy classification backward projection technique and an adaptive iterative stopping strategy are used to enhance the denoising effect.Experimental results show that the proposed algorithm obtains better objective indicators and visual inspection.