Semiautomatic Road Extraction From VHR Images Based on Multiscale and Spectral Angle in Case of Earthquake
| Autor: | Idrissa Coulibaly, Michèle St-Jacques, Richard Lepage, Nicolas Spiric |
|---|---|
| Rok vydání: | 2018 |
| Předmět: |
Atmospheric Science
010504 meteorology & atmospheric sciences Computer science business.industry Feature extraction 0211 other engineering and technologies 02 engineering and technology 01 natural sciences Disaster area Kernel (image processing) False detection High spatial resolution Computer vision False alarm Artificial intelligence Computers in Earth Sciences business Image resolution Spectral angle 021101 geological & geomatics engineering 0105 earth and related environmental sciences |
| Zdroj: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing. 11:238-248 |
| ISSN: | 2151-1535 1939-1404 |
| DOI: | 10.1109/jstars.2017.2760282 |
| Popis: | Road extraction offers great potential for research initiatives because of the complexity due to its great topological variability. The use of remote sensing imagery to accomplish this mapping is an interesting option. Indeed, satellite images can be acquired shortly after the event, and cover a large area of territory. We hope to produce a mapping of the present facilities from very high resolution images shortly after a disaster. This availability of very high spatial resolution images brings added value to the study in urban areas and their mapping. Increasing the spatial resolution generates noise, which makes extraction difficult, especially in the event of an earthquake in an urban context. This problem increases false alarm rates and generally affects the performance of road extraction algorithms in detecting linear features used to locate and extract roads on such images. During major disasters, short deadlines demand an effective response in terms of updating the mapping of affected areas. Our aim is to improve the road extraction quality after adaptation of Lowe's scale-invariant features transform descriptors jointly with spectral angle algorithms. An illustration is performed on three high-resolution images, respectively, representing a rural, suburban, and urban disaster area, captured by the Quickbird satellite. Our approach significantly reduces the false detection rate and shows an increase in overall quality of up to nearly 30% in some cases as compared to what obtain in the literature. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|