High Repetition Rate Underwater Range-gated Image Adaptive Denoising Algorithm Research

Abstract

High repetition rate underwater image is vulnerable to underwater scattering noise and ICCD device noise vulnerable. After decomposed by wavelet transform, the fact is obtain that the noise approximately obey generally gaussian distribution in low frequency sub-band images and the noise approximately obey generally salt and pepper distribution in high frequency sub-band images. A new denoising algorithm is proposed as AWGNLM in wavelet filed which uses general non-local means as low frequency and adaptive median filter as high frequency. The experimental results show that compared with other algorithm the proposed algorithm can not only get higher signal noise rate and lower speckle index, but also get much better image information.