Automated Building Damage Assessment Using Remote-Sensing Imagery

Cyclone damage detection on building structures from pre- and post-satellite images using wavelet based pattern recognition

“…The standard deviations of pixel radiance of the three visible layers, i.e., red, green and blue layers, and also the standard deviation and entropy of the gray scale intensity obtained by the wavelet feature extraction method on all the damaged buildings, shows almost a linear increase as the percentage area of roof damage obtained by automatic texture wavelet analysis increases. But conventional features extracted by the change detection method (Womble et al, 2007) show a scattered relation with the percentage area of roof damage obtained by automatic Canny edge detection, although it showed a linearly increasing relation with the manually classified damage scale, RS Scales (Womble et al, 2007).…”

“…(Sabareesh et al, 2006;Douglas and Pillay, 2005) either spatial domain information of image data (or time domain for any 1-dimensional data) or frequency (spectral) domain information is available at an instant, not both spatial and spectral information together. In the field of wind damage detection from satellite images, a statistical feature extraction technique has been used (Womble et al, 2007;Womble, 2005). Hence, there is a chance of losing major information.…”

“…But the introduction of high-resolution satellite imageries has created a breakthrough in identification of disaster affected areas for rescue purposes as well as reconstruction of wind disaster damaged buildings. In Womble et al (2007) and Womble (2005), computational analysis was done for building damage detection mainly using statistical analysis of the histogram of the satellite image pixel radiance value in four spectral bands. Damage to building structures is rated by ground truth data into an RS-Scale (Remote Sensing Scale) table.…”

Automated Detection of Damaged Areas after Hurricane Sandy using Aerial Color Images

“…The first method is called "change detection" and is based on identifying changes on an object at different times (Singh, 1989). Several change detection techniques were employed in previous works for detecting changes in buildings from pre-and post-disaster images (Brunner, et al, 2010, Myint, et al, 2008, Radhika, et al, 2013, Turker and Cetinkaya, 2005, Womble, et al, 2007. For example, tornado affected areas were identified using remote-sensing images and three change detection approaches, including the Principal Component Analysis (PCA), Image Differencing, and Object-oriented Classification (Myint, et al, 2008).…”

Use of post-storm images for automated tornado-borne debris path identification using texture-wavelet analysis

“…Research on cyclone damage to buildings was estimated from pre-and post-storm satellite imagery by Womble et al (2007), Radhika et al (2011). Much research have also been conducted for other types of natural disasters such as earthquakes using aerial imagery by Matsuoka and Yamazaki (2000), Hasegawa et al (2000), Mitomi et al (2001), Sumer and Turker (2004) and Ozisik (2004).…”