Checksum-aided video error correction for real-time, streaming and broadcast video applications

点击翻译按钮获取中文摘要

In the recent years, we have witnessed an impressively rapid deployment of video devices, applications and systems, and the growth of video will only intensify in the coming years with emerging technologies such as 4K/8K Ultra HD formats, video devices for remote monitoring and control of machinery, transport and healthcare, virtual and augmented reality, and other devices resulting from the emerging Internet of Things. However, transmission errors over unreliable networks (e.g. wireless and satellite) cause significant visual impairments or dramatically reduce the effectiveness of networks (e.g. by causing retransmissions). Such errors significantly degrade the user's quality of experience (QoE) in most real-time and near real-time video applications. ******The main objective of this research program is to investigate low-complexity solutions to the problem of video error correction, which can be realistically integrated into communication systems. Specifically, the applicant and his team will study a new application layer video error correction approach, which has the potential to fully repair damaged video packets under reasonable error conditions. This approach consists in exploiting the protocol's checksum information, currently used only to detect errors, to correct errors. ******The applicant and his team will achieve the main objective through three research activities. First, in the context of real-time video, they will investigate how the checksum information can be exploited in a list decoding approach, which attempts to decode various candidate corrected video packets until a valid (decodable) packet is found, to dramatically reduce the number of candidates to consider and, therefore, the computational complexity. Second, in the context of near-real-time video streaming, they will investigate how the checksum can be used to limit the number of retransmissions by combining the information of a damaged packet with its retransmitted damaged version(s) to produce a repaired packet. Finally, in the context of video broadcasting using forward error correction (FEC) (redundant information), they will investigate how the checksum can be used in conjunction with FEC to produce a repaired packet. The technologies conceived in this research program will be validated on current as well as future video standards. ******The advances provided by this research will bring significant improvements in the quality and efficiency of video services and enable the emergence of unprecedented media experiences. As for their previous projects, Canada will strongly benefit from the intellectual property and commercialization of these technologies and from the training of highly qualified personal in the fast growing field of video processing and applications.