Structural Depth Estimation Via Depth Maps Of 3 Dimensional Videos

Abstract

3-Dimensional (3D) Video Quality Assessment (VQA) has been an important area for researchers working in this area. The reason is there isn't any well accepted and standardized VQA method for 3 Dimensional (3D) as for 2-Dimensional (2D) video. Depth perception assessment (DPE) is the most critical part of 3D VQA because of visual realism. Subjective tests are currently in use for the 3D VQA because there aren't any 3D VQA algorithms for measuring this perception accepted by researchers in literature. Subjective tests are not ergonomic methods from the stand point of time and cost. Therefore, it is quite important to develop objective 3D VQA metrics for predicting the depth perception of users. The VQA algorithms developed without using a reference video is called No-Reference (NR) metrics in literature and they are considered efficient compared to the other metrics. In this study, Depth Maps (DM) in 2D+depth based 3D videos are utilized to measure Structural Average Depth (SAD) in a NR manner. The results of this study presents that the YOD algorithm can be considered as a part of a 3D VQA metric assessing the depth perception and approved by researchers.