采用改进梯度互信息和粒子群优化算法的红外与可见光图像配准算法

Registration algorithm of infrared and visible images based on improved gradient normalized mutual information and particle swarm optimization

针对传统互信息图像配准容易产生局部极值,以及传统梯度互信息配准方法计算量大等问题,在互信息和梯度方法基础上构建了一种改进的梯度互信息方法,该方法直接统计梯度图像的互信息,有效地将图像梯度信息和灰度信息结合起来,不仅保证了配准精度,而且较传统梯度互信息方法减少了计算量。在参量优化的过程中,针对传统粒子群优化算法易陷入局部极值的缺点,提出了改进的粒子群优化算法,该算法在传统粒子群优化算法基础上引入混沌优化思想和遗传算法中的杂交思想,不仅能够有效抑制局部极值,而且加快了收敛速度。多种红外与可见光图像配准实验结果证明,文中提出的算法能够有效提高配准精度和速度。

In order to solve the problem that using classical mutual information measure in infrared and visible images registration may suffer from local extremum,and large amount of calculation by using classical gradient normalized mutual information measure,an improved gradient normalized mutual information measure based on classical gradient and mutual information was proposed,which counted mutual information of gradient image directly to combine image gradient information with gray information effectively.Compared with classical gradient normalized mutual information measure,the new measure could improve registration precision and reduce computation cost.During the optimization of parameters,for the defect of sinking into local extremum for classic particle swarm optimization algorithm,the improved particle swarm optimization algorithm was proposed,which included chaos optimization idea and hybridization idea in genetic algorithm.The improved particle swarm optimization algorithm can restrain local extremum and accelerate convergence.Experimental results demonstrate that this new algorithm can achieve high registration efficiency.