激光与光电子学进展, 2017, 54 (11): 112801, 网络出版: 2017-11-17   

灾后高分辨率遥感影像的地震型滑坡信息自动提取算法研究 下载: 698次

Automatic Extraction Algorithm of Seismic Landslide Information Based on After-Calamity High-Resolution Remote Sensing Image
作者单位
1 中国科学院遥感与数字地球研究所数字地球重点实验室, 北京 100094
2 中国科学院大学, 北京 100049
摘要
针对地震滑坡灾害应急响应的高时效性要求,提出了一种基于灾后高分辨率遥感影像的地震滑坡体自动提取算法。该算法综合利用了高分辨率遥感影像的光谱、形状和纹理等特征,基于多特征阈值分层次逐步剔除干扰地物,实现了地震滑坡体的自动提取。涉及到的特征参量阈值均采用改进的Otsu算法自动确定。在利用2008年汶川地震后ADS40航空遥感影像自动提取滑坡的实验中,所提算法的滑坡个数正确提取率超过70%,面积正确提取率超过80%。对于10000 row×10000 column的ADS40影像,算法执行时间低于1 min。相较于传统的人机交互目视解译方法,该算法的自动化程度高、滑坡提取速度快,滑坡识别精度可以满足地震灾害应急要求。
Abstract
Aiming at the high efficiency requirement of extraction seismic landslide, we propose an automatic extraction algorithm of seismic landslide based on after-calamity high-resolution remote sensing image. The algorithm utilizes the spectrum, shape, and texture features of seismic landslide of the remote sensing image to remove the disturbed features based on the analytic hierarchy process, and realizes the automatic extraction of the seismic landslide. All the feature parameters are computed automatically by the improved Otsu algorithm. In a test using high-resolution aerial remote sensing data acquired by ADS40 image after earthquake Wenchuan in 2008, the experiments demonstrate that more than 70% landslides are correctly detected by the proposed method. Furthermore, the area accuracy is more than 80% and the extraction time is less than one minute for the ADS40 image with 10000 rows and 10000 columns data. Compared with traditional visual interpretation of human-computer interaction, the proposed algorithm has high degree of automation and landslide extraction speed, and the landslide recognition accuracy can meet the earthquake disaster emergency requirements.

闫琦, 李慧, 荆林海, 唐韵玮, 丁海峰. 灾后高分辨率遥感影像的地震型滑坡信息自动提取算法研究[J]. 激光与光电子学进展, 2017, 54(11): 112801. Yan Qi, Li Hui, Jing Linhai, Tang Yunwei, Ding Haifeng. Automatic Extraction Algorithm of Seismic Landslide Information Based on After-Calamity High-Resolution Remote Sensing Image[J]. Laser & Optoelectronics Progress, 2017, 54(11): 112801.

本文已被 3 篇论文引用
被引统计数据来源于中国光学期刊网
引用该论文: TXT   |   EndNote

相关论文

加载中...

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!