A content-aware error resiliency scheme in video transmission over wireless channels

Compared with wire-line network, wireless network has the characteristics such as severe bandwidth constraints, high error rates and time varying nature. Video transmission in wireless environment is a challenging task to enable reliable, flexible, adaptive, and robust communication. Traditional error control schemes consider videos as low-level bit streams, ignoring the underlying visual content. The content-aware framework is based on the recognition of strong correlation among video content, required network resources, and resulting video quality. If video content is taken into consideration during error resilience process, the bandwidth can be utilized efficiently, and "focus of attention" (FOA) can be protected hierarchically, the reconstructed video can satisfy the user's demand of information maximally. The content-aware error resiliency schemes proposed in this paper divide the video content into three hierarchies according to human's attention sequence, then use content feature to classify the content of each shot into a set of shot classes, and then associate each shot class with representative error resiliency model. The corresponding model is tested to be the best one for the shot class. During video communication, shots are classified and transmitted with the associated error resiliency scheme. As a result, the most valuable information is achieved at the minim cost.