一种基于多粒子群协同进化的高光谱图像波段选择与分类方法

针对高光谱图像分类过程中数据波段多以及信息冗余量大引起的处理速度慢及Hughes现象等问题,提出了一种基于多粒子协同进化算法进行高光谱图像自动波段选择与分类的方法:使用多粒子群协同进化算法搜索特征子集,对粒子群优化算法进行改进,定义新的位置和速度的更新策略,并以支持向量机为分类器,同时对特征子集和SVM核函数参数进行优化。在协同搜索过程中,引入遗传算法改善粒子群优化的"早熟"收敛问题,构建了一种新的MPSO-SVM(Multiple particle swarm optimization-SVM)分类模型。对高光谱遥感图像的实验结果表明:MPSO-SVM方法不仅能有效地压缩光谱的特征维数,得到最佳的波段组合,还能得到最优的SVM参数,达到较好的分类效果,提高分类精度。

Band Selection and Classification for Hyperspectral Image Based on Multiple Particle Swarm Cooperative Optimization

The huge increase of hyperspectral data dimensionality and information redundancy has brought high computational cost as well as the risk of over-fitting when classification is performed.We presented an automatic band selection and SVM classification method based on a novel wrapper multiple particle swarm cooperative optimization-SVM model (MPSO-SVM),which uses multi-particle swarm algorithm to search the feature subset,and improves the PSO by the new update strategy of position and velocity.In the process of cooperative optimization,we improved the premature convergence of PSO by introducing genetic algorithm.The MPSO-SVM model optimizes both the band subset and SVM kernel parameters simultaneously.The experimental results on hyperspectral image demonstrate that MPSO-SVM can select the best band combination and the optimal SVM parameters,and improve the classification accuracy significantly.