Thermal infrared images are characterized by pixel grayscale, and the target usually has a higher grayscale than the background, so it is easy to be detected. The texture features of visible light images are mainly represented by gradients, and the gradient size (such as edges) provides detailed information for the scene. Therefore, fused images are expected to preserve the dominant intensity distribution in infrared images and the gradient changes in visible light images.

 

Other total variation models retain the characteristic information of thermal infrared images and visible light images at the same time, and the fusion result is essentially a high-resolution thermal infrared image that clearly highlights the foreground target, retaining the maximum or approximately maximum amount of information.