论文部分内容阅读
针对传统遥感变化检测算法中差值影像构造方法不容易提取弱变化信息,以及变化阈值需要人工干预的不足,提出一种基于奇异值分解(SVD)和最大类间方差法(OTSU)阈值分割的遥感影像变化检测算法。首先计算两期遥感影像的多波段差值影像,对其进行矢量化后所构造矩阵进行奇异值分解,并计算差值影像的变化强度,选取奇异值分解主分量投影和变化强度作为表征两期影像变化的特征量;然后通过最大类间方差法对上述两个特征量进行阈值分割,得到变化检测结果;最后通过对比实验以及精度验证,证明了该变化检测方法相比传统方法能够更精确地提取出变化信息,而且能够自适应获取分割阈值。
Aiming at the problem that it is not easy to extract the weak change information in the method of constructing differential image in traditional remote sensing change detection algorithm, and the threshold of human intervention needs to be changed, a threshold segmentation algorithm based on singular value decomposition (SVD) and maximum inter-class variance (OTSU) Remote sensing image change detection algorithm. Firstly, the multi-band difference image of two remote sensing images is calculated, the matrix constructed by vectorization is singularly-valued decomposed, and the variation intensity of the difference image is calculated. The principal component projection and intensity of singular value decomposition Then the variance of the above two features is divided by the maximum class variance method to get the change detection result. Finally, the comparison experiment and the accuracy verification show that the change detection method can be more accurate than the traditional method Extract the change information, and can adaptively obtain the segmentation threshold.