| In all of earthquakes happened since the founding of PRC, the Wenchuan Earthquake which occurred in Sichuan province of China on May12th of2008then spread to more than10provinces and400cities quickly, is a serious catastrophe whose influence area should be most widely, relief work most difficult, and destruction force also most powerful. In order to effectively minimize the losses of disaster, accurately develop the relief measures, practically improve the efficiency of emergency rescue, the salvation headquarters must rapidly obtain the earthquake information then swiftly assess the disaster damage. In the case of communication interruption and ground transport disruption, the Remote Sensing (RS) becomes an important means for earthquake intelligence obtaining, emergency response and post-earthquake disaster assessment. However, the Satellite Remote Sensing subject to certain restriction such as spatial resolution and run cycle when we use it to get the disaster information. Meantime, the limitation also happened to the traditional Manned aerial photography. The airport and weather conditions will seriously affect the usage of Manned aerial photography, and there may be some potential dangers on pilots when flying. As a supplementary resort to SRS and Manned aerial photography, the Unmanned Aerial Vehicle (UAV) should have some distinct advantages which can be generalized as follow:real-time, flexible, high image resolution, low cost, and so on. Due to that the aircraft is driven by virtual computer not real man, the UAV can be used in high-risk areas then rapidly get the photos of earthquake area. This is very useful for disaster survey, losses assessment, and measures development. Therefore, in the field about emergency relief, the UAV has broad application prospect and development space.In the process of earthquake emergency treatment, data timeliness requirement is very high whose specific content will be as follow:the rescue team should decode seas of date obtained from aerial photography as soon as possible and accordingly ensure the accuracy of these interpreted maps about earthquake religions. Owing to the tiny size and light weight, the unmanned aerial vehicles are especially vulnerable to the effects of airflows hence deviate from the established air route. These peculiarities just mentioned above will increase the rotation angle of UAV camera and the curvature of flight line, also make the image heading and lateral overlap irregular, the image amplitude small and the image number many, the baseline short and irregular, etc. Besides, those traits make UAV can not use the professional aerial camera and attitude measuring instrument, and similarly can not timely and accurately get high-precision control points. Hence all the characteristics of UAV have brought a lot of formidable challenges to the in-house data processing. Based on the above shortcomings of UAV, we’d better not use the ordinary Digital Photogrammetry to deal with these aerial photographs. In fact, we will encounter many problems such as indicators over run, the iterative computation difficult to converge, if we forced to use this measure. Moreover, this misuse will cause a series of troubles to the later image processing, such as the processing time is too long, and therefore this method cannot meet the requirements of the emergency rescue. Unlike the means mentioned above, the basic method in this paper is fast splicing these UAV images without ground control based on SIFT algorithm, then geometric correcting these processed photos according to the topographic map, in order to obtain some valuable graphics about earthquake timely and widely. Although this means for splicing images is not based on strict collinear equation yet, the practice has proved that it has a strong real-time and fully meet the needs of disaster emergency response.When using UAV images, information extraction is a very critical step in the process of emergency rescue. In this paper, we make use of the object-oriented classification technique which is a remote sensing images classification method suitable for the features of UAV such as high-resolution, in order to fully excavate kinds of intelligences from these high-resolution images thus improve the efficiency and accuracy of these extractions. In fully considering the various features such as geometry, shape, semantics, texture, and topological relations, etc, we can classify these specified objects by multi-scale segmentation method. The first related example in this paper is that extracts earthquake information from those earthquake building images in Zundao Town of Mianzhu County. To extract information rapidly, we should classify these UAV images whose resolution is accurate up to0.5m, according to spectrum, shape, texture on a single scale. Another pointed case is the collapsed damage in Wenchuan earthquake. For this same purpose, we use the hierarchical classification method to classify these images whose resolution is accurate up to0.42m, according to spectrum, shape, texture and context at different levels. Generally speaking, in order to achieve the objective of rapid information extraction from earthquake damage, it is advisable to analyze some issues according to actual situation. These specific approaches are as follow:we must use fuzzy rules to distinguish specified UAV images if the objects in images have distinctive features; on the contrary, the comprehensive method combined the nearest classifier with the fuzzy rules will be used if the objects are mixed category hard to discern. In short, this provide lots of complementary data for earthquake investigation and rapid disaster assessment. |
PDF Full Text Request