In order to solve storage and computation cost problems for the traditional whole sampling image fusion algorithms, a new method of infrared and visible light image fusion is put forward based on compressed sensing (CS) theory. Nonsubsampled shearlet ...
In order to solve storage and computation cost problems for the traditional whole sampling image fusion algorithms, a new method of infrared and visible light image fusion is put forward based on compressed sensing (CS) theory. Nonsubsampled shearlet transform (NSST) is introduced as the sparse transform. Compressed sensing is applied to fuse the high frequency subbands decomposed by NSST. The high frequency coefficients are compressed for measured values which are fused by the rules of spatial frequency weighting. Regional energy together with regional standard deviation guides the fusion of the low frequency subband. Finally, the fused image is gained through inverse NSST. The experimental results show that high-quality fused images can be obtained with only one layer NSST. The fused image quality is better than the several traditional fusion algorithms based on compressed sensing.