Dissertation > Excellent graduate degree dissertation topics show

DEM Generated from SPOT Stereo Pairs and Its Uncertainty Research

Author: CaoZhaoFeng
Tutor: PanSongQing；XuHanWei
School: Hohai University
Course: Photogrammetry and Remote Sensing
Keywords: SPOT DEM Accuracy Uncertainty
CLC: P231.5
Type: Master's thesis
Year: 2006
Downloads: 631
Quote: 3
Read: Download Dissertation

Abstract

At the present time, the research of the uncertainty of spatial data is very significant. As an important component of the basic establishment of spatial data, the uncertainty of DEM is also very important. As a digital expression of earth surface, DEM has a widespread application. There are many methods to generate DEM. But the traditional methods take a lot of time and effort. And they are inclined to generate the problem of input error. They have a property of hysteresis quality. With the continuous development of aerospace technology and aerial technology and the continuous increasement the performance of sensors, DEM can be got by photogrammetric surveying after stereo pairs have been got by RS technology.This paper extracts DEM using SPOT images and VirtuoZo Workstation which is popular in our country at present. There are errors and uncertainties inevitably when extracting DEM. This paper makes use of Monte Carlo method to simulate the uncertainty of extracted DEM using Jiangning as an experimental area.The control points are collected from topographical map of 1:50000 when generating DEM. The collection and distribution of the control points are very important for the accuracy of DEM. Using VirtuoZo to generate DEM includes error as follows: the accuracy of topographical map isn’t high, thus accuracy of the control points is low. The resolution of the image is 10m, which is not advantage to the collection of the control points. The area of Jiangning is complex and cracked, which will influence the computation of elevation. Then it will influence the accuracy of DEM. The selection of matching window will influence the accuracy of DEM. There are personal error, system error, random error and so on.Before evaluating the accuracy of DEM, this paper disposes the blunder of DEM.It uses the the method of detection and rejection of the blunder which is based on the grid DEM. The result indicates that the number of blunder points is larger where the surface is higher, while the number of blunder points is smaller where the surface is flater.This paper makes use of checkpoint method and profile method to evaluate the accuracy of DEM. When using checkpoint method to evaluate the accuracy of DEM, the extracted DEM is compared with DEM of 1:50000. The author selects 20 checkpoints, and computes that RMSE is 11.86m. The author makes a conclusion: the accuracy of extracted DEM is better where the surface is flat. When using profile method to evaluate the accuracy of DEM, the author draws 8, 9, 11 section lines in Xaspect, Y aspect and arbitrary aspect respectively. And the author computes RMSE of each section line and the mean of RMSE. The author finds that the mean of RMSE of section line of different aspect is similar. And they are similar with RMSE of 20 checkpoints. They are satisfied with the accuracy of DEM of 1:100000 fundamentally. Using VirtuoZo to extract DEM is satisfied with the accuracy of DEM of 1:100000 for Jiangning whose configuration is complex and crushing, then other regions should be also satisfied with these requirements.This paper makes use of Monte Carlo method to simulate the uncertainty of DEM. It emphasizes on the generation of random field and the determination of N. By comparing various methods of the generation of random field, this paper uses random method to generate random field. This can gurantee to simulate the uncertainty of DEM in the worst case. The paper uses two kinds of RMSE to simulate the uncertainty of DEM. When RMSE is 11.86m, N is 150. When RMSE is 15.85m,N is also 150. The paper also paints the error map between simulated DEM and DEM of 1:50000, and finds that the accuracy of simulated DEM is better where the surface is flatter. That is, the simulated DEM is close to the true value where the surface is flatter.This paper improves Monte Carlo method. Considering the relation of mean and slope, the paper fits the linear function between them. By using this function as the mean of random field, the paper considers the influence of land complexity on the simulation of DEM. The paper also compares the method which has been improved with the method which is not improved, and finds that the improved method can simulate DEM whose accuracy is higher.

Related Dissertations

CLC: > Astronomy,Earth Sciences > Surveying and Mapping > Photogrammetry and Surveying, Mapping and Remote Sensing > Aerial Photogrammetry > Digital Mapping
© 2012 www.DissertationTopic.Net Mobile