Bayesian Compressive Sensing Based on Importance Models

To solve the problem that all row signals use the same reconstruction algorithm, a type of Bayesian compressive sensing based on importance models is proposed, which reconstructs more important signals firstly even if losing some unimportant signals. Compared to Bayesian compressive sensing whose performances is not well when sampling ratio is lower, the proposed algorithms can improve reconstruction quality effectively. The importance models include two processes, one is judging whether the signal is important and the other is how to reconstruct important signals better. In this paper, the improved reconstruction algorithm is based on sparse important signal and assigning measures by important weights. The two algorithms give priority to the more important column coefficient signals in the reconstruction process. The experimental results show that the proposed algorithms have better reconstruction effect than the traditional Bayesian compressive sensing, and especially, the performance of reconstruction algorithm based on assigning measures by important weights is improved obviously when the sampling rate is relatively low. Copyright © 2013 IFSA.