基于自适应变异的变步长天牛须算法及其在图像配准中的应用

针对现有遥感技术缺乏光谱信息而导致图像失真度高、成像模糊等问题, 文章提出基于倾斜摄影技术提高遥感图像分辨率的方法。传统的天牛须算法(beetle antennae search algorithm, BAS) 在处理图像匹配时, 虽然具有参数少、收敛速度快、易于实现等优点, 但是精度有限。因此以步长指数衰减的方式进行变步长搜索, 同时借鉴粒子群算法寻优策略, 引入天牛左右须历史最佳位置作为下一步解的搜索参考位置, 以此提高解的搜索速度。与传统的天牛须智能优化算法相比, 该文提出的基于自适应变异的变步长天牛须算法求解精度更高, 通过实验验证了该算法的有效性。

In response to the problems of high image distortion and blurred images due to the lack of spectral information in the existing remote sensing technology, this paper proposes a strategy to improve the resolution of remote sensing image based on oblique photography technology. The traditional beetle antennae search algorithm (BAS) has the advantages of fewer parameters, fast convergence speed, and easy implementation when dealing with image matching, but the accuracy is limited. In this paper, in the basic resolution of the step size, the variable step size search is carried out by making the step size decay exponentially. At the same time, referring to the particle swarm optimization strategy, the best position of the left and right whisker of BAS is introduced as the reference position of the next solution search to improve the solution search speed. Compared with the traditional BAS, the improved BAS proposed in this paper has higher solving accuracy. Finally, the effectiveness of the method is verified by experiments.