Infrared and visual image fusion through infrared feature extraction and visual information preservation

Abstract

The ideal fusion of the infrared image and visual image should integrate the important bright features of the infrared image, and preserve much original visual information of the visual image. To achieve this purpose, we propose a simple, fast yet effective infrared and visual image fusion algorithm through infrared feature extraction and visual information preservation. Firstly, we take advantage of quadtree decomposition and Bézier interpolation to reconstruct the infrared background. Secondly, the infrared bright features are extracted by subtracting the reconstructed background from the infrared image and then refined by reducing the redundant background information. To inhibit the over-exposure problem, the refined infrared features are adaptively suppressed and then added on the visual image to achieve the final fusion image. In this way, the fusion image could not only reveal the invisible but important infrared objects by integrating the infrared bright features, but also show good visual quality by preserving much original visual information. Experiments performed on the commonly used image sets validate that the proposed algorithm outperforms several representative image fusion algorithms in most of the cases.