Cube-based perceptual weighted Kronecker Compressive Sensing: Can we avoid non-visible redundancies acquisition?

Compressive sensing approach directly avoids the acquisition of statistical redundancies of a signal. However, perceptual redundancies of images and videos due to the human eye sensitivity are not considered so far. Besides, an effective sampling scheme is needed to multidimensional signal reconstruction using a low number of measurements to avoid all redundancies. In this paper, along with the Kronecker structure of the sampling matrix we design various weighting matrices based on the spatio-temporal contrast sensitivity function to avoid acquisition of non-visible redundancies. Moreover, inspired by the block-based compressive sensing, we divide a group of pictures in a video sequence into cubes. Hence, the size of measurement and sparsifying basis matrices are reduced and the reconstruction algorithm can be implemented in parallel. We further show that our simple linear sampling approach can be competitive with motion compensation method. Simulation results verify that our proposed method notably outperforms the other state-of-the-art methods.