Automatic detection of temporal synchronization mismatches between the stereoscopic channels for stereo 3D videos

Abstract

This paper presents an efficient approach to estimate temporal synchronization distortions in 3D sequences, based on measuring motion consistencies of object feature points in depth planes. Errors in post-processing steps or exposure setups of stereo cameras can cause temporal asynchronization between stereoscopic channels. This kind of 3D artifacts lead to poor 3D experience for viewers. Due to the fact that the synchronization mismatch between stereoscopic views is normally slight, the perceptibility of temporal synchronization mismatches is first analyzed since distortion is sometimes not noticeable. Experimentally, this framework introduces 1, 2 or 3 frames of slight temporal deviations to simulate the synchronization mismatch between stereoscopic channels. The experimental results show that the proposed framework has a high accuracy for detecting the synchronization mismatch of a given 3D sequence at the frame and the shot levels.