A comparative analysis of pixel-level image fusion based on sparse representation

Along with the development of wavelet transform (WT) technique, sparse representation as a new tool also gradually lead to the researchers´ wide interests. In our study, a sparse representation based unified image fusion model is introduced. Under this fusion model, we compare several fusion results of combining different overcomplete dictionaries and sparse coefficients, and eventually find a best combination. Extensive experiment is implemented on two source images, and results illustrate the proposed fusion model is very effective. Furthermore, we also find an optimal combination of dictionary and sparse coefficient for pixel-level image fusion.