Comparing objective visual quality impairment detection in 2D and 3D video sequences

Abstract

Thanks to the availability of 3D-capable televisions and bluray players, 3D content is made accessible in the home. Recently, an extension of the H.264/AVC video coding standard has been defined for encoding 3D video content. This extension, called Multiview Video Coding, allows inter-view prediction resulting in a better compression efficiency. However, due to these inter-view dependencies impairments in one view caused by e.g. packet losses can lead to degradations in other views. Research has already been conducted towards estimating packet loss visibility in H.264/AVC encoded sequences. In this paper, we investigate the possibility of using an existing decision tree-based classifier for estimating impairment visibility in 3D MVC encoded sequences. Our results show that, in the case of losing entire pictures, it is possible to estimate packet loss visibility in 3D MVC encoded sequences with a high accuracy by only taking into account a limited number of parameters.