Improved Video Denoising Algorithm Based on Spatial-Temporal Combination

Abstract

An algorithm of video denoising based on spatial-temporal combination is proposed to improve the video quality. This algorithm can adaptively distinguish the still regions from the motion regions of video frames by bilateral motion detection of multiple frames, where temporal bilateral Kalman filtering is applied to the still regions and spatial bilateral adaptive Nonlocal means (ANL) filtering is applied to the motion regions. Experimental results show that by making full use of the spatial-temporal information of video, the proposed algorithm can significantly improve the peak signal-to-noise ratio (PSNR) and the subjective video quality without movement ghosting. Besides, the speed of video denoising can be accelerated by bilateral filtering.