Edge-Based Reduced-Reference Quality Metric for 3-D Video Compression and Transmission

Abstract

3-D video applications are delivered over a range of different transmission systems. In order to provide demanding customers with a better service over unreliable communication channels, compression and transmission system parameters can be changed “on the fly.” For interactive 3-D video services, video compression can be adapted (e.g., it can be made more robust and/or rate adaptive) based on the quality measured at the receiver. It has been shown that measuring the (3-D) video quality at the receiver-side, and using this information as a feedback to fine tune the system parameters, will result in improved performance in such systems. However, measuring 3-D video quality using Full-Reference (FR) quality metrics is not feasible due to the need of the original 3-D video sequence at the receiver-side for comparison. Therefore, this paper proposes a Reduced-Reference (RR) quality metric for color plus depth 3-D video compression and transmission, using the extracted edge information of color plus depth map 3-D video. This work is motivated by the fact that the edges/contours of the depth map can represent different depth levels and this can be considered for measuring structural degradations. Since depth map boundaries are also coincident with the corresponding color image object boundaries, edge information of the color image and of the depth map is compared to obtain a quality index (structural degradation) for the corresponding color image sequence. The performance of the method is evaluated for different compression ratios and network conditions. The proposed method achieves good results compared to its counterpart FR quality metric, with a lower overhead for side-information.