The Application Of Noise Estimation Algorithm In Image Denoising Of Side-scan Sonar Images

With the implement of the national strategy target of building a maritime power,Various marine engineering and scientific research are widely carried out,which increases the demand for marine geomorphic information rapidly.Side-scan sonar is widely used in the field of seabed detection because of its long range of action,wide coverage,strong penetration and large data collection.However,due to the complex and varied working environment of the side-scan sonar,the effect of noise on side-scan sonar image is serious,and the image information is lost or difficult to interpret.Therefore,the de-noising of the side scan sonar image is of great significance for obtaining the information of the submarine detection,and it is the basis of the subsequent image processing.Most of the existing image de-noising algorithms are based on optical images,most of which are good at dealing with the noise of optical images.The noise of the side-scan sonar image is multiplicative noise,which is different from the additive noise of the general optical image.Therefore,most of the optical image de-noising algorithms can't be applied directly to the side-scan sonar image de-noising.The side-scan sonar images usually have high brightness wave front and low brightness back surface.Side-scan sonar image,as an acoustic image,has its own unique image characteristics,and is decided by the working mode and working environment of the side-scan sonar system.Moreover,the existing de-noising algorithms are lack of targeted analysis and processing of side-scan sonar image characteristics and noise types,and no attention is paid to the coupling effect of side-scan sonar image signal and noise.In this paper,the image de-noising algorithm and noise estimation algorithm for side-scan sonar image are studied.The main work content and contribution are as follows:(1)This paper discusses the research significance of side-scan sonar image de-noising.The measurement principle and system composition of the side-scan sonar system are introduced briefly.Then this paper gives de-noising models based on different noise types,and analyzes the source of the noise of the side-scan sonar image,and gives the noise model of the side-scan sonar image according to its noise characteristics.the domestic and foreign existing image de-noising algorithm is introduced in detail,and introduces the algorithm of de-noising algorithm commonly used in the prior parameters of noise variance,briefly introduced the research status of domestic and foreign noise estimation algorithm.This paper summarizes and points out the shortcomings of the existing de-noising algorithms application to the de-noising of side-scan sonar images.(2)The noise spillover phenomenon of side scan sonar images is studied.The basic characteristics of side-scan sonar image are introduced,based on the type of side-scan sonar image noise,pointed out that the coupling effect between the side-scan sonar image and noise,pointed out that the coupling effect between the side-scan sonar image and noise,pointed out the phenomenon that the side-scan sonar image gray value overflows the noise gray scale,when the image is under the influence of the noise.The adverse effects of the overflow phenomenon brings to the noise estimation and de-noising are introduced.(3)The noise variance estimation algorithm based on the weak texture block is proposed based on the side scan sonar transform image,and the BM3 D de-noising algorithm is improved according to the calculated noise variance.Based on the noise feature and noise spillover of the side scan sonar image,on the basis of the transform image,through the iterative calculation of the threshold of the weak texture block and the variance of the texture block,the image block under the influence of the noise is selected to estimate the noise.The calculation results are applied to the selection of the BM3 D de-noising threshold,which improved the performance of BM3 D algorithm.The simulated image and the actual side scan sonar image de-noising experiments are carried out based on the improved BM3 D de-noising algorithm.The experimental results show that the algorithm has excellent de-noising effect and the texture is restored in the high brightness region,and the de-noising effect of this algorithm is obviously superior to other algorithms in the low luminance region.The performance of proposed algorithm on processing of gray value overflow is obviously superior to other algorithms.(4)A variation coefficient noise estimation algorithm based on iterative computation is proposed based on the side scan sonar image and the BM3 D denoising algorithm is improved according to the calculated variation coefficient.Taking the multiplicative noise of the side sonar image as the research background,the image block which is affected by the noise spillover is eliminated by the coefficient of variation of the image block,and the selection of the image block is controlled by the coefficient of variation of the image block.The noise estimation result is used as the threshold parameter of the BM3 D de-noising algorithm,which improved the performance of BM3 D de-noising algorithm.The simulated image and the actual side scan sonar image de-noising experiments are carried out based on the improved BM3 D de-noising algorithm.The experimental results show that the variation coefficient noise estimation algorithm using iterative settlement,which remove the influence of the noise spillover,has a good representation to the noise distribution,and the noise estimation results are more accurate.The de-noising performance of the BM3 D algorithm based on the variation coefficient is better than the other two algorithms,and the details remain well.