Robust Video Denoising for Better Subjective Evaluation

Abstract

Video denoising is a pre-processing step of many high level video processing applications since it can considerably enhance the perceived video quality, increase compression effectiveness, facilitate transmission bandwidth reduction and improve the performance of the subsequent higher level task such as feature extraction, object detection, prediction, coding and object tracking. In this paper, a new scheme for video denoising based on spatial filtering is proposed which synthesizes the blurring caused by averaging filter and sharpening due to high boost filter for effective objective and subjective video quality enhancement. The main drawback of averaging filter is that it gives a blurring effect and de-emphasizes the fine details from an image such as edges and texture. This blurring artefact is compensated through highboost filtering operation by selecting an appropriate value of the highboost weight factor intended for better video quality and superior HVS performance. Experimental results show that the proposed method achieves better performance of noise removal with much reduced blurring and fine detail preservation in terms of improved objective and subjective video quality.