The Analysis of Reconstruction Efficiency with Compressive Sensing in Different K-Spaces

Compressive sensing is a potential technology for lossy image compression. With a given quality, we may represent an image with a few significant coefficients in the sparse domain. According to the sparse modeling theories, we may randomly sense a few number of measurements in a transform domain and later reconstruct the sparse representation. Typically the sensing domain is a low-complexity transform domain and the computation complexity lies on the reconstruction phase. In this paper, the linear and nonlinear compressive sensing approaches are briefly introduced. A few experiments are performed based on the nonlinear approach. Both 2D-DFT and 2D-DCT sensing domains are included to show their effects to the reconstruction quality. The simulation shows that the two domains produce comparable results if the proper comparison condition is considered. Some directions of revising the reconstruction process is also discussed in this paper.