Digital

Wissenschaftliche Publikation

A Hybrid Wavelet and Temporal Fusion Algorithm for Film and Video Denoising

Publikation aus Digital

DI Hannes Fassold, DI Peter Schallauer

Berlin Das Symposium für Wissenschaftler, 10/2015

Abstract:

Noise is an impairment which often occurs in both flm and digital video and severely degrades the viewing experience of the content. In this work, we propose a two-phase algorithm for flm an video denoising. In the first phase, the concept of semi-local shrinkage functions is used to exectively separate noise from image structure. In the second phase, we show how to fuse the result images of the first phase in a way which is robust to motion-compensation errors. A quantitative evaluation of the proposed algorithm on a realistic test dataset with noise of different coarseness and magnitude shows that the proposed method delivers better results than the video denoising method CVBM3D.