论文部分内容阅读
为了进一步提高多时相遥感图像变化检测的精度,本文提出了一种将Shearlet变换与核主成分分析(kernel principal component analysis,KPCA)相结合用于遥感图像变化检测的算法.首先利用Shearlet变换的多尺度、多方向和各向异性等特点,对遥感图像进行多尺度分解,然后对分解后的数据进行核主成分分析,再进行Shearlet反变换得到含变化信息的图像,最后对该图像利用模糊局部信息C均值(fuzzy local information c-means,FLICM)聚类算法进行分割,实现遥感图像的变化检测.大量试验结果表明,与基于主成分分析(principal component analysis,PCA)、基于KPCA、基于小波变换和PCA 3种变化检测算法相比,本文算法能有效地分离出变化信息,得到更准确的变化检测图像,具有更高的变化检测精度,且对背景有较强的鲁棒性,同时也减少了计算复杂度.
In order to further improve the accuracy of multi-temporal remote sensing image detection, this paper presents an algorithm that combines Shearlet transform and kernel principal component analysis (KPCA) for remote sensing image change detection.First, Scale, multi-direction and anisotropy, the multi-scale decomposition of remote sensing images is carried out, and then the principal component analysis of the decomposed data is performed. Then the Shearlet inverse transform is used to obtain the image with change information. Finally, the fuzzy local (FLICM) clustering algorithm to realize the change detection of remote sensing images.Many experimental results show that, based on principal component analysis (PCA), based on KPCA, based on wavelet transform Compared with PCA three kinds of change detection algorithms, the proposed algorithm can effectively separate the change information, obtain more accurate change detection images, have higher change detection precision and strong robustness to the background, meanwhile reduce The computational complexity.