Robust Sparse Representation Combined With Adaptive PCNN for Multifocus Image Fusion

Multifocus image fusion utilizes the clear parts of multiple images from the same scene to generate a new image, which includes more information about the scene than any of the individual source images. In this paper, a multifocus image fusion method based on robust sparse representation (RSR) and an adaptive pulse-coupled neural network (PCNN) is presented. Each source image is first decomposed with RSR to obtain a sparse coefficients matrix and a residuals matrix. Second, the spatial frequency of the residuals matrix is calculated as the motivation for PCNN neurons, and a salience map of the source image is proposed as an adaptive linking strength for the PCNN. The initial decision map is acquired by comparing the ignition frequency maps of the source images obtained through PCNN. Then, the final decision map is achieved through morphological opening and closing operations. Finally, the fused image is obtained by employing a weighted fusion rule. Experimental results indicate that the proposed method is effective and better than other existing popular fusion methods regarding both objective and subjective evaluations.