Infrared and visible image fusion based on FUDPCNN and gravitational force operator

Kunal Bansal, Vinay Kumar, Chirag Agrawal, Siddharth Tripathy, Samarth Bandhu Khera, Garima Singh, Chinmaya Panigrahy, Ayan Seal

Abstract: Infrared and visible image fusion merges the salient details of an infrared and its respective visible image to create a formidable image more suitable for surveillance, image enhancement, object detection, and remote sensing. This paper presents a multi-scale transform-based infrared and visible image fusion method in the non-subsampled contourlet transform domain. The proposed method utilizes a new fast unit-linking dual-channel pulse coupled neural network (FUDPCNN) model. The low-pass sub-bands are fused by a new gravitational force operator-based mechanism. On the other hand, the internal activities of the proposed FUDPCNN are applied to get the fused high-pass sub-bands. The effectiveness of the FUDPCNN is shown by comparing it with practiced PCNN models. Moreover, the competitiveness of the gravitational force operator-based rule is described using other low-pass rules. The workability of the proposed method is shown by comparing its fusion outcomes on well-known infrared-visible image pairs with the nine existing approaches using eight objective metrics.