A New Prediction Structure for Efficient MV-HEVC based Light Field Video Compression

Light Field imaging has emerged as a technology that enables the capture of images and video with richer information. Captured content is composed of numerous views aligned in both horizontal and vertical directions providing full parallax, offering light intensity and directional information, but at the same time significantly increasing bandwidth requirements. Several multi-view coding methods have attempted to tackle this problem. However, these approaches do not fully assess the intricacies that are found in light field content. This paper proposes a prediction structure for coding light field content using the MV-HEVC standard, exploiting the inter-view correlations in two directions along with the high similarity between views around the center of each frame. Experimental results show BD-rate gains up to 38% compared to an existing state-of-the-art method.