A State of Art Methodology for Real-Time Surveillance Video Denoising

摘要

Video de-noising is one of the most essential tasks in video processing. Noise elimination and video reconstruction are still remaining as a challenging problem in this area. In this paper a modified two dimensional block based least mean square algorithm (TDBLMS) proposed to reduce the noise in the video, which can increasingly improve the image quality. The processing includes updating the adaptive weight matrix to filter noisy frame block by block and achieve higher peak signal to noise ratio. The performance of the modified TDBLMS approach compares with existing TDBLMS algorithm and gives better PSNR and SSIM results.