Infrared image fusion enhancement method based on a gradient confidence variation model

The invention discloses an infrared image fusion enhancement method based on a gradient confidence variation model, and mainly solves the loss of some source images in the conventional total-variationfusion method. The method comprises steps of: 1, obtaining an infrared image u and a visible light image v; 2, constructing the multi-scale structure tensor weighted gradients of the infrared image uand the visible light image v; 3, using an average image to initialize the fused image of the first iteration and setting the maximum iteration number N; 4, constructing the brightness fidelity itemand the gradient fidelity item of the fused image; 5, constructing the energy functional composed of brightness fidelity item and gradient fidelity item; 6, solving an energy functional minimization problem, obtaining the fusion image of the (T+1) iteration; 7, repeating the step 4 to step 6 until the number of iterations reaches N, and outputting a fusion result. The method has good target retention performance, rich detai