The Application Research Of Rapid Access Information In The Chengdu Plain By UAV Survey

The weather of Chengdu plain is the main of multi-cloud. The conventional aerial and satellite mapping technology limited by the relatively large, low altitude unmanned aerial vehicle(UAV) remote sensing technology would be overcome the difficulty of weather to obtain information on the Chengdu plain, and it has advantages of flexible and high efficiency. But the instability of UAV leads to a serious distortion of their images which caused severe restrictions on the UAV in the survey mapping application. How to complete UAV images mosaic quickly and how to reduce the distortion of images to improve mapping accuracy by UAV images have very important significance for the development of UAV.In this paper, a village in Chengdu plain of reconstruction after the 5.12 earthquake was selected as an experimental area, low altitude UAV remote sensing technology was used to obtain image data of experimental area. Combined with the UAV images have the characteristics of high resolution and large distortion, a rapid UAV image processing system was developed by the Visual C# and Matlab as a platform. The system achieve automatic UAV image mosaic, automatic extraction of control points, and the topographic map and 4D (DEM, DOM, DLG, DRG) were produced by used domestic and foreign image processing software. A strong technical support and a wealth of mapping data were provided for the reconstruction of the experimental area.The main contents are as follows:1. The domestic and foreign UAV technologies in the application of rapid information access were discussed and the current statuses of the study of rapid mosaic images were analyzed.2. The conventional methods of aerial survey data processing and UAV data processing were summarized and compared. LPS was selected as a platform, the large scale topographic map, DEM, the production of three-dimensional landscape of experimental area were completed, and the variety mapping method of efficiency, accuracy and cost were compared and analyzed.3. Automatic mosaic UAV images was realized by used Visual C# and Matlab as a platform and improved SIFT algorithm was introduced.4. Automatic extraction of control points was realized based on the UAV attitude parameters (POS data) and collinear equation conditions.5. The method of object-oriented image classification was used in UAV image classification, and the experimental area landuse information was obtained and classification accuracy was assess.