基于多样性变异的QPSO算法的遥感图像分类

遥感图像分类是遥感领域研究的热点问题之一。结合量子粒子群优化(QPSO)算法和多样性变异的机制提出了一种新的高光谱遥感图像分类算法。在遥感图像分类过程中,采用无监督分类,图像中每个像素点到聚类中心的高斯距离作为分类标准,使用QPSO算法进行聚类中心的优化,在聚类过程中使用多样性变异机制防止QPSO算法早熟收敛,使分类结果达到最优化。在遥感图像上所做的实验表明:此分类算法具有较好的搜索速度和收敛精度,能有效寻找和优化最佳聚类中心,是一种有效、可行的遥感图像分类方法。

Classification of multispectral remote sensing image based on QPSO and diversity-mutation

The classification of remote sensing images is one of the most important issues in remote sensing today. This paper presents a novel classification algorithm for multispectral remote sensing images based on the quantum-behaved particle swarm optimization(QPSO) algorithm and diversity-mutation. To classify remote sensing images, we adopted unsupervised classification, and used the Gaussian distance function between the image pixels and the cluster centers as the classification standard. We used the QPSO algorithm to optimize the cluster centers. For clustering, we propose diversity-mutation to prevent premature convergence of the QPSO algorithm to optimize the classification results. The experimental results show that the proposed algorithm not only has better search speed, but also has higher convergence precision, and searches and optimizes the best cluster center more efficiently. Therefore, we conclude that the algorithm is effective and feasible.