Change detection based on adaptive fusion of multiple features
-
摘要: 针对传统的变化检测算法主要依赖像斑的光谱信息,未能有效地利用影像多特征检测优势的问题,基于面向对象的分析思想,提出一种多特征融合的遥感影像变化检测算法.首先,以多尺度分割的影像对象为基础,统计各对象的颜色直方图和边缘直线梯度直方图;然后,利用推土机距离计算不同时期对象之间的颜色距离和边缘直线特征距离,采用自适应加权方法将颜色距离和边缘直线特征距离组合构建对象的异质性;最后,采用直方图曲率分析获得像斑的变化检测结果.实验结果表明,该方法能够充分融合颜色和边缘直线特征,提高变化检测的精度.Abstract: In view of the fact that the traditional change detection algorithm mainly depends on the spectral information and fails to effectively use image feature detection advantage,the authors put forward a multi-feature fusion of remote sensing image change detection algorithm. First, color histogram and edge histogram of gradient image object with multi-scale segmentation is statistically analyzed based on the calculation of each object. Then, the object color distance and edge linear characteristics distance between different periods are calculated by using the earth mover's distance method;the adaptive weighted method is used to combine color distance and edge linear characteristics distance so as to construct object heterogeneity. Finally, the images change detection results are analyzed by using histogram curvature. The experimental results show that the method can fully fuse the color and edge line features and improve the accuracy of detection.
-
-
计量
- 文章访问数: 707
- PDF下载数: 30
- 施引文献: 0
下载: