Font Size: a A A

Research On De-Noising Algorithm Based On Sonar Image Speckle Noises

Posted on:2020-03-12Degree:MasterType:Thesis
Country:ChinaCandidate:L J ZhuFull Text:PDF
GTID:2392330602458412Subject:Control Science and Engineering
Abstract/Summary:PDF Full Text Request
Images generated in the complex and changeable underwater environments will be disturbed by many interferences.These interferences are main factors affecting the quality of sonar images because they can make the images fuzzy and lead to low resolution and serious noise.As a result,sonar images have different features from optical images.In order to better use sonar images,it is very important to carry out denoising processing to sonar images.Moreover,sonar images have different characteristics from optical images,so the noise processing algorithm for optical image cannot be directly applied to denoising process of sonar images.Therefore,it is very significant to build denoising algorithm for sonar images.This paper researches the following contents accoridng to features of sonar images:(1)This paper did some research and analysis of noise types of the sonar image and process noise types that have great influence on sonar images.The research found that the multiplicative speckle noise whose amplitude value obeys Rayleigh distribution will have relatively serious impacts on the sonar images.This paper gets a simplified method for multiplicative speckle noise model according to the existing noise model construction method.(2)According to features of sonar images,this paper applies 3D block matching image de-noising algorithm(BM3D)to the de-noising processing of sonar images and use total variation de-noising algorithm to improve the filtering processing parts of BM3D algorithm.Experimental verification shows that the improved algorithm has good de-noising ability to multiplicative speckle noises in sonar images.At the same time,it can reserve more texture information and marginal information.(3)Because of the good performance of total variation and wavelet transform in image processing,this paper proposes that the non-convex regularized total variation-wavelet de-noising algorithm which can well reduce signal noises should be used to reduce multiplicative speckle noises of sonar imag]]es and that the dual-tree complex wavelet transform which can well process singular points of images should be used to improve wavelets of the original algorithm so as to obtain an improved algorithm which is more suitable for de-noising processing of sonar images.Second,in order to ensure that the improved algorithm can have a better de-noising ability to sonar images,this paper carries out improved to the parameter function.With the aid of the simulation verification,the improved algorithm can better adapt to the de-noising processing of sonar images.This algorithm has both good de-noising abilities and good texture information and edge-preserving ability.
Keywords/Search Tags:Double-tree complex wavelet, Total variation algorithm, Sonar image denoising, Non-convex regularization term
PDF Full Text Request
Related items