Distributed Compressive Video Sensing by Adaptive Allocation of Frame Measuring Rate

The traditional distributed compressive video sensing usually samples each non-key frame with the same measuring rate. This method usually leads to an inferior reconstruction quality because it does not take into account the correlation-difference among different frames. To address this problem, we formulate a new allocation model for the measuring rate of each non-key frame firstly .Then, an effectively adaptive allocation algorithm which utilizes the inter-frame correlation is proposed according to the new allocation model. The proposed algorithm allocates the measuring rate of each non-key frame by exploiting its contribution to the reconstruction of current GOP (Group of Pictures). The experiment results show that, in the same sampling rate, the reconstructed video quality is improved when taking the proposed adaptive allocation approach into the scheme.