Efficient Measurement of Stereoscopic 3D Video Content Issues

摘要

The paper presents metrics to estimate a number of spatial and temporal parameters relevant for stereoscopic 3D video content. Based mainly on view differences and disparity, the aim of these metrics is to check for common issues with 3D content that might make viewers uncomfortable. The algorithms are designed for high computational efficiency to permit real-time video content analysis, and are shown to be robust to common video impairments.