Font Size: a A A

Research On Key Technology Of Terrain Three-dimensional Reconstruction Based On High Resolution Remote Sensing Image

Posted on:2018-11-13Degree:MasterType:Thesis
Country:ChinaCandidate:D ZhuFull Text:PDF
GTID:2348330512489077Subject:Electromagnetic field and microwave technology
Abstract/Summary:
With the rapid development of satellite remote sensing technology, high-resolution remote sensing images have become one of the essential tools for discovering the geography. High-resolution satellites can acquire images in short cycle without the limitation of geographical area and weather condition, which also ensure the accuracy of measurement. As a new source of data for mapping research, high resolution images have great research and application value.This paper carried out theoretical research on three-dimension reconstruction and verified it’s practicability using the high resolution stereoscopic pair of ZY-3. The main contents are listed below.(1) This paper analyzed the mechanism and methods of popular sensor imaging models for remote sensing satellites. Considering the universality and accuracy of 3D reconstruction, the ration function model (RFM) is selected for further studies. This article researched the mechanism of rational function model. The two solution methods,including the terrain unrelated model and the terrain related model of the coefficients in rational function model, are discussed.In process of solving rational polynomial coefficients (RPC), the correlation between each other often lead to morbid equation, which means greater error. To solve this problem,this article improves the terrain correlation method based on collinearity analysis to calculate the RPC. Experimental results show that the proposed method can weaken the morbidity of the equation and improve the accuracy of the RFM in image localization.(2) Taking the stereoscopic pair of ZY-3 as an example, the problem of image matching based on stereo image pair is studied. The SIFT feature matching algorithm are studied and implemented using the images of ZY-3. Based on the preliminary feature matching of SIFT algorithm, the improved RANSAC algorithm is used to remove incorrect matching feature points. The experiment results shows that this improved algorithm removes the wrong matching points, and reduces the time cost of 36.76% with almost accuracy.(3) This paper analyzes the principle and method of 3D reconstruction based on stereo image pair. Using the stereoscopic pair of ZY-3, the RPC and the matching points,digital elevation model (DEM) of the coverage area is generated. A test for evaluating quality of DEM is conducted. The results show that the root mean square error (RMSE)of DEM is 4.24m. Compared with the traditional terrain correlation method and image matching method of ENVI, the algorithm based on collinearity analysis and improved image feature matching algorithm designed in this paper have positive effects.(4) In the tenn of 3D visualization, this article analyzes the structure of World Wind Java and uses its software development kit (SDK) to implement the 3D visualization of ZY-3 high resolution image and the DEM. On this basis, the functions about loading 3D models, geospatial measurement and terrain profile are realized by using Java programing language.
Keywords/Search Tags:3D reconstruction, digital elevation model (DEM), rational function model, image matching, 3D visualization
Related items