Study On The Geometric Processing Key Technologies Of Aerial And Low-Altitude Remote Sensing Images For Emergency Response

Using aerial/low-altitude remote sensing platform to acquire image is a very effective kind of method to obtain spatial data, it has an advantage in the field of emergency response and three-dimensional urban construction. Images obtained in emergency response situation are very different from aerial images in traditional regular situation. The former one often has larger rotation angle and higher degree of overlap, but irregular. The UAV low altitude remote sensing system also has many shortcomings, such as distortion, small size and image redundancy. The mentioned problems can lead to some troubles in the sequent image matching, orientation and aerial triangulation. It is urgent and necessary to solve those problems.In order to solve the existing problems of the aerial/low-altitude photogrammetry in emergency response, this paper considered the common problems between the manned aero photogrammetry and UAV low altitude photogrammetry. UAV low altitude image is the main research object in this paper, the contains mainly include as follows:1. Review on the geometry processing of the aerial/low-altitude remote sensing images. It includes the pre-processing, the extraction of the tie points and sparse bundle adjustment. This work provides the foundation for automatic aerial triangulation.2. Quantitative analysis of the existing five feature extraction operator including DOG、MSER、Harlap、Heslap and SFOP in the automatic aerial triangulation, which provide proof to the automatic extraction of the tie point.3. Considering bundle adjustment of large amounts of data, the building methods of the normal equation and the transformed normal equation are elaborated, what’s more, the matrix structure are analyzed in detail. Corresponding indexical structure and storage algorithm is also introduced, which can substantially improve the efficiency of bundle adjustment. These technologies provide guarantees to the fast processing of UAV image data.4. According to the large rotation angle existing in aerial/low-level images obtained in emergency response situation, an absolute orientation method considering the large rotation and scale differences by using of the GPS data of the POS data (Position and orientation system) is proposed to orient the whole block after the free network is generated. After that, bundle block adjustment is employed refine the parameters to provide measurable results. This part of work can provide the product with geocoding.Simulated data and a great number of aerial Images and UAV Images are used to performed the method proposed in the paper. The results reveal that the proposed method can complete image orientation without control points, and then measurable coordinates of the three-dimensional points are obtained to provide useful product for emergency response.