Abstract
Addressing transmission errors in underwater acoustic channels is a key challenge for the real-time video communication between an autonomous underwater vehicle and a surface station. In this paper, we propose an error-resilient video compression technique based on hybrid multiple descriptions and redundant pictures to overcome impact of packet loss in underwater acoustic transmission. Video sequences are split into two descriptions with selective redundant pictures in order to achieve a balance between coding efficiency and error resiliency. Experiments with underwater video sequences are presented to assess the performance of the proposed approach under various packet loss rates, in comparison to state-of-the-art single and multiple description coding techniques.
