针对已有红外和可见光图像融合算法存在的复杂度高、实时性差、输入参量多和融合模型受先验知识影响等问题, 提出了红外和可见光图像泊松融合算法.算法采用绝对值最大规则融合源图像的梯度得到泊松融合图像的期望梯度, 将平均融合图像分解, 分别作为迭代的初始值和边界条件, 利用雅可比迭代求解泊松方程重建泊松融合图像.通过实验分别确定了最佳迭代次数以满足高质量和高实时性两种融合需求.与经典融合算法相比, 该算法在质量优先或速度优先的图像融合应用方面性能更优.

Aiming at the problems of high complexity, poor real-time performance, too many input parameters and the fusion model affected by the prior knowledge on existing infrared and visible image fusion algorithms, a Poisson fusion algorithm of infrared and visible images is proposed. In the algorithm, the expected gradient of the Poisson fused image is obtained by fusing the gradient of source images with the absolute maximum rule. The average fused image is decomposed into initial values and boundary conditions of the iteration. The Poisson fused image is reconstructed by solving the Poisson equation with the Jacobi iterative solution. For the two fusion needs of high quality performance and high real-time performance, optimal numbers of iterations are determined by experiments. In comparison with classical fusion algorithms, the proposed algorithm has batter performance for both quality-first fusion applications and speed-first fusion applications.