Compressive sensing based simultaneous fusion and compression of multi-focus images using learned dictionary

In this paper, we present a framework of fusion and compression of multi-focus images using learned dictionary. A single dictionary, learned from a set of natural images is used to initially fuse the multi-focus images. Using the same dictionary as the basis matrix, the fused coefficients are compressed using compressive sensing theory. Recovery of the fused image using the compressively sensed measurements is carried out at the receiver end using well known Sl0 recovery algorithm. Fusion and compression is thus achieved simultaneously using a single learned dictionary. Experiments on multi-focus images show the effectiveness of the proposed method in fusing and compressing the images concurrently. Simulation results also verify that the proposed method outperforms some of the existing compression methods especially at lower sampling rates.