Infrared and visible image fusion based on object extraction and adaptive pulse coupled neural network via non-subsampled Shearlet transform

Taking into account the imaging characteristics of the infrared and visual images and the insufficient information content of the fused images, combined with the benefits of non-subsampled Shearlet transform and adaptive pulse coupled neural network, a kind of infrared and visual image fusion algorithm based on object extraction and adaptive pulse coupled neural network via non-subsampled Shearlet transform is proposed, which fuses the object region and background respectively. First, the object regions from infrared image are segmented by region growing method. Then, non-subsampled Shearlet transform is utilized for multiscale geometric decomposition of the source images, the object regions and background region are fused by different rules. The high frequency sub-band coefficients of background region are selected by using the global coupling and pulse synchronization of the adaptive pulse coupled neural network, and finally the fused image is reconstructed by the inverse non-subsampled Shearlet transform. Experimental results demonstrate that the proposed fusion algorithm outperforms typical wavelet-based, Contourlet-based and non-subsampled Contourlet-based fusion algorithms in terms of qualitative and quantitative evaluations.