Post Disaster Information Extraction of Damaged Houses from Multi-Scale Image Segmentation
Special Issues Editor (Nottingham Tent University, United Kingdom (Great Britain))
Advances in photographic sensor technology for high resolution images has provided opportunities for the application of remote sensing to address the limitations of traditional multi-scale segmentation based on image regions. The paper provides a new method of multi-scale segmentation which is combined with edge detection and regional feature detection. We use nonlinear filtering to segment, and then extract successfully the information of a house damaged by earthquake. We unite the edge detection method using the Canny operator to increase the confidence of an advanced regional feature detection method of multi-scale segmentation using the Mean Shift operator. The new method limits effectively the phenomenon of both over and under segmentation, while maintaining the space structure of the object. The total confidence value of Kappa reached 0.9623 in the experiment and the time of multi-scale segmentation shortened to less than 20 seconds while processing two images to provide post disaster information.
Journal: International Journal of Simulation: Systems, Science & Technology, IJSSST V17
Published: Jul 14, 2016