In order to improve the image quality of lidar fusion images,an improved regularization algorithm was designed.The algorithm introduces sparse representation weighted coding into the regularization algorithm in the form of parameters,which is used to reasonably assign the weights of noise suppression types and improve the robustness of the algorithm.The objective function including iteration coefficient h and regular term parameter λ is deduced,and the specific process steps of algorithm implementation are given.The experiment optimizes the fusion image containing shot noise.The results show that as h increases,the image quality improves,and when h reaches 40,the image quality tends to be stable.When λ∈(0,0.5),the image quality increases as the increases of λ,and when λ∈(0.5,1.0),the image quality decreases as the increases of λ.It can be seen that this algorithm achieves the optimal solution when h=40,λ=0.5,and its timeliness is better than the traditional algorithm.It can be seen that this algorithm has certain application value in enhancing the image quality of fused images.