Fusion of infrared and visible images based on a hybrid decomposition via the guided and Gaussian filters

Abstract

Due to the characteristics of infrared and visible images, in the process of image fusion, the two kinds of heterogeneous image information will interfere with each other, which lead to the fusion image become blurred, confused and suffers contrast degradation. Therefore, a novel fusion method for infrared and visible images based on a hybrid decomposition via guided and Gaussian filters is proposed in this paper. Firstly, guided and Gaussian filters are used to achieve a hybrid decomposition for infrared and visible images, then the small-scale texture details, large-scale edges and coarse image information of the input images can be obtained. The subinformation is then merged, and the fusion weights are determined by the large-scale edges which can represent the features of the infrared image. In this way, the prominent infrared image information can be injected into the visible image. In the end, the fused image can be reconstructed by assembling the merged sub-information. Experimental results show that the proposed is better than the typical multi-scale decomposition based image fusion algorithms both in subjective and objective evaluation.