Remote-sensing images fusion by compressed sensing in contourlet transform domain

The compressive fusion by block-based compressed sensing (BCS) produces a low computation cost, but suffers from low quality in recovery since BCS sampling lacks in global feature. Therefore, a new compressive fusion with the contourlet-transform-based BCS (CTBCS) is proposed. Firstly, the CTBCS is implemented with block-size and subrate dependent on decomposition-level. Then the compressive samplings are fused by rule of linear weighting. Finally, the fused image is reconstructed by the iterative thresholding projection (ITP) algorithm with removal of blocking artifacts. Field test shows the CTBCS achieves better compressive fusion than that by BCS does. With better consideration of global feature, the CTBCS fusion simplifies fusion decision and provides better compressive fusion for big images of remote-sensing.