摘要
提出一种基于频域显著性(FDS)方法和极限学习机(ELM)方法进行遥感影像变化检测的方法.首先,对利用变化矢量分析方法(CVA)获取不同时相遥感影像的光谱特征差异图及纹理特征(灰度共生矩阵法)差异图进行融合获得差异影像(DI);然后,利用频域显著性方法获取差异影像的显著性图,采用模糊c均值(FCM)聚类算法对显著性图选取阈值得到的粗变化检测图进行预分类(变化像素、未变化像素、待定像素);最后,从光谱及纹理特征影像上提取变化像素和未变化像素的邻域特征作为可靠样本进行ELM训练,并利用训练好的ELM分类器对粗变化检测图进行变化检测,得到最终的变化检测图.通过对高分辨率遥感影像数据实验结果表明本方法的变化检测精度及性能优于其他对比方法.
A method based on frequency domain significance(FDS)method and extreme learning machine(ELM)was proposed for remote sensing image change detection.First,different images(DI)of the multitemporal remote sensing images was obtained by fusing the spectral feature difference map and the texture feature(grey level co-occurrence matrix)difference map obtained by change vector analysis(CVA).Then,the DI saliency map was obtained by the frequency domain significance method,and the coarse change detection map obtained by selecting a threshold for the DI saliency map was pre-classified(changed pixels,unchanged pixels,undetermined pixels)by the fuzzy c-means(FCM)clustering algorithm.Finally,the neighborhood features extracted from spectral and texture feature images of the changed pixels and the unchanged pixels in the coarse change map were used as reliable samples for ELM training.The trained ELM classifier was then used to perform change detection on the coarse change detection map so as to obtain the final change detection map.Experiments on high-resolution remote sensing image data show that the change detection accuracy and performance of the proposed method is better than that of the contrast methods.
作者
王昶
张永生
韩世静
于英
WANG Chang;ZHANG Yongsheng;HAN Shijing;YU Ying(School of Civil Engineering,University of Science and Technology Liaoning,Anshan 114051,Liaoning China;Institute of Surveying and Mapping,Information Engineering University,Zhengzhou 450001,China;School of Natural Resources and Surveying,Nanning Normal University,Nanning 530001,China)
出处
《华中科技大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2020年第5期19-24,共6页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金资助项目(41501482)。