Onboard real-time compression of launch vehicle telemetry images

Image information plays a very important role in fault diagnosis for launch vehicles. A video surveillance sub-system based on the present wireless communication system is designed and its corresponding on-board encoder is developed. To prevent from video-quality degradation and the losing frames problem caused by unexpected scene changes during flight tests, an optimal encoder framework is presented which could detect scene changes while encoding, and only I frames are encoded whenever the scenes change rapidly. Two algorithms are also proposed for reducing the computation cost of H.264. The first is a novel all-zero DCT block detecting algorithm which could improve detection efficiency without any additional computational cost by combining sum of absolute difference with adaptive detection thresholds in table. The second is a fast mode-selection algorithm which could early terminate the mode selecting operation by regarding all-zero DCT block as the best one. Experiments show that the proposed algorithms can achieve approximately a 30%-45% computational saving and almost the same PSNR compared with JM13.2, and the encoder can fulfill the requirement of real-time encoding (25f/s). Moreover, both lost frames and video-quality degradation do not occur when unexpected scene changes appear.