Study On The Noise Of Laser Underwater Imaging System And Enhancement Technology

The technology of underwater laser imaging has very important practical value at ocean exploitation, but its imagery is usually affected by speckle noise. Although synchronous scanning technology and range-gated technology are mostly applied, it can hardly guarantee the perfect processing results. A large number of background noises are brought to underwater laser image for the reasons of absorption and scattering of water, which is also the cause of noises and low contrast of the underwater images. In order to improve the effect of the images, digital image processing techniques are introduced.This paper introduces several technology of green-blue laser underwater target detection developed in recent years such as range gated imaging system, synchronous scanning, polarized laser imaging, streak tube imaging lidar. The respective imaging principle, characteristics and update status are presented.To analysis the back scatter of a imaging system, the optical back scatter of water arriving at a detector is calculated, a model of optical back scatter of the underwater imaging system is given. The theoretical calculations are programmed, and figures of overlap coefficient and optical back scatter plotted based on typical system parameters. Which noise on the dominating status is very important for filtering algorithm, so, the statistic characteristics such as probability curve and gray mean variance retioetc. In the imagery disturbed by speckle and some other gaussian noises are analyzed firstly, and then the different statistic characteristics affected by the two kinds of noises respectively are found. Based on these characteristics, the images of underwater laser imaging system are discussed, and then the results show that the most imagery noise is speckle.This paper displays the powerful ability of MATLAB in image processing with underwater laser image processing. Median filtering is a non-linear image processing method that can preserve image edge while removing noise. In light of computing the speckle indexes, PSNR and NMSE of the processed images, their ability of suppressing speckle noise is compared.