The Research Of Earthquack Information Extraction And Assessment Based On Object-oriented Technology With Remotely-sensed Data

Earthquake, being bursty, transitory and powerful is thought as one of the greatest natural disasters in the world, it poses a grievous threat to human's lives and properties. China is a country who suffers badly from the earthquake disaster. But up till now, it is impossible to forecast the exactly time, place and magnitude of an earthquake. Under this situation, the efficient way to reduce the earthquake loss and protect human lives is to do well in protecting against and emitipung earthquake disasters project pre-earthquake and to obtain the earthquake information at the first time and finish the evaluation work quickly so as to organize the anti-quake and relief work as soon as possible. Remote sensing characterized with synchronism, timeliness, comprehensiveness and economy of information provides a new technology to earthquake information extraction and assessment. With the development of high-resolution, hyper-spectral, multi-platform and multi-date remote sensing technology, the method of earthquake information extraction has changed fundamentally and step into a new stage.This paper, aimed at quick, precise and efficient, did research in application of High-resolution remote sensing on earthquake information extraction and assessment in order to explore a suite of procedure and way so as to accumulate some valuable experience and make a contribution to this field. There are two contents of this paper:On the one hand, the research of earthquake information extraction, including:the preprocessing of multi-source remote sensing data, the information extraction based on change detection technology (change detection method, image segment and earthquake information identification and extraction), the slope information extraction with pixel-oriented and object-oriented method, the dilapidated buildings extraction with pixel-oriented and object-oriented method; On the other hand, earthquake information evaluation, this paper brings forward the concept of earthquake disaster damage risk, assessment model as well as criterion and proceeds the assessment of the study area.The main research findings and innovations are summarized as follows:(1) The study put forward the preprocessing technique flow of multi-sources remote sensing data; there are three stages of geometric correction. Besides the geometric correction, the technique flow can also accomplish the image mosaicking, image subdivision, image fusion and radiometric enhancement. The result shows that the procedure feasible and reliable enough to meet the high-quality requirement of remote sensing data for earthquake information change detection.(2) A new change detection method of multi-sources data based on MNF/ICA theory was brought forward. Compared to other traditional change detection ways, this method can not only reduce the appearance of noise by overcoming the high-order relevance of remote sensing data but also detach change information from noise so as to concentrate and highlight the change information of different data. The outcome suggests the MNF/ICA method enhances the accuracy and effect of change information detection obviously.(3) The research put up with an image segment algorithm called two-threshold fuzzy recognition algorithm. This algorithm can eliminate the noise and some individual land cover classes mixed in change information to some extent as compared to one-threshold segmental algorithm. The segmental image shows more precise result.(4) By analyzing the landslide characters of study area, the present study advanced a quick-and-effective way of landslide extraction with NDVI and slope information and validated with pixel-oriented and object-oriented methods. It is shown that the object-oriented way can successfully avoid the phenomenon of NDVI-differential bright noise caused by the spectral diversity of high-resolution remote sensing data and make the result more scientific and accurate.(5) A pixel-oriented method on the basis of the multi-layer regional segment idea which embodies the thoughts of layer and segment from object-oriented was applied to collapsed buildings extraction. As compared to traditional way, it is an innovative experiment. When it comes to object-oriented method, the multi-scale segment algorithm is applied on order to build up three-layer hierarchy. Then by analyzing the spectrum, texture, shape, location and context of individual classes in different layer, the fuzzy determined rule system is established for earthquake collapsed buildings extraction. We compared with the two results from three aspects which are principle, visual impact and precision assessment. It is depicted that the object-oriented analyzing method has good potential and prospect of application to earthquake information extraction as the overall accuracy of 90.3% as compared to 76.84% of pixel-oriented method.(6) The elevation, slope, hydrographic net, land use, vegetation cover and special factor were selected as the indicators to build up the estimation system. Then the AHP was applied to set up EDRI index in order to assess the earthquake damage risk of study area. As the factor of earthquake information is introduced to the assessment indicators and plays an important role in determination matrix, the evaluation result is reliable, scientific and sensitive to earthquake information.