研究目的
To introduce the technology of texture-wavelet analysis for detection of roof damages due to cyclones and tornadoes from close range remote sensing imageries and to quantify the Degree of Damage (DoD) by calculating the percentage of damaged portion of the building roofs.
研究成果
The study successfully demonstrated the use of texture-wavelet analysis for rapid and accurate identification of roof-damaged buildings from post-disaster images alone, achieving a high classification accuracy and correlation with visual inspection data. This method aids in prioritizing emergency aid and reconstruction efforts.
研究不足
The method relies on the availability of post-disaster images and may face challenges in areas with dense vegetation or complex urban structures that obscure roof damages.
1：Experimental Design and Method Selection:
The methodology involves texture-wavelet analysis on post-disaster remote sensing images to detect and quantify roof damages.
2：Sample Selection and Data Sources:
Close range satellite imagery of Punta Gorda, Florida USA, hit by hurricane 'Charley' and aerial image data of Saroma town in Hokkaido, Japan, hit by a tornado.
3：List of Experimental Equipment and Materials:
QuickBird satellite imagery and aerial images provided by Shin Engineering Consultants Co. Ltd.
4：Experimental Procedures and Operational Workflow:
Identification of disaster location, segmentation of buildings, wavelet feature extraction, SVM classification, and texture-wavelet analysis for DoD determination.
5：Data Analysis Methods:
Wavelet decomposition and reconstruction, feature extraction (standard deviation and entropy), SVM classification, and calculation of percentage area of damage.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
