Dissertation > Excellent graduate degree dissertation topics show

Study on Extracting Informations of Mining Exploitation Area Base on Multi-source Remote Sensing Images

Author: WangShaoHua
Tutor: TianShuFang
School: Chinese Geology University (Beijing)
Course: Resources and Environmental Remote Sensing
Keywords: Mining area of Multiscale segmentation Object-oriented Maximum Likelihood Classification Accuracy Assessment
CLC: P627
Type: Master's thesis
Year: 2011
Downloads: 123
Quote: 0
Read: Download Dissertation

Abstract

Mining economy has become one of the pillar industries in China, but there are still many problems. Remote sensing technology is successfully applied to the investigation and monitoring of the order of the mine development, improve the supervision of the mine administration department, But when the face of multi-resolution, vast amounts of remote sensing data, how efficient and accurate extraction of thematic information is still difficult to overcome One of the problems. Based classification pixel classification accuracy is not high, \The man-machine interactive interpretation method extraction accuracy, but requires rich interpretation of experience and expertise, and there is the lack of long life cycle and low efficiency. In this paper, Miyun District of Beijing in the partially open-air mining as the research object, GeoEye-1, RapdiEye Landsat-5 three kinds of satellite image data try mine development covers an area of ??information extraction methods. Classification method to the classification of the object-oriented and traditional pixel-based maximum likelihood classification method based on the accuracy of the results of the two classifications were compared and analyzed, the following results: 1) system summarizes the traditional classification methods and object-oriented Classification of principle of the method. Elaborate on the object-oriented classification process, the method of key technologies such as image segmentation, information extraction research. Comparative study of five different fusion methods) for GeoEye-1 image brightness information from the information entropy, correlation coefficient, the relative deviation of four evaluation evaluation that IHS fusion method is better. 3) for medium-resolution RapidEye data the optimal segmentation technology research through comparative analysis of multiple scale segmentation image object, determine 40,30,20 three split scale build object hierarchy extraction in the classification system mining area of ??information. 4) according to the multiscale segmentation constitute the object hierarchy to set up a classification rule base. Object-oriented classification method to make full use of the image spectrum, space, texture, context information, using the combination of a variety of characteristics mine covers an area of ??information of the same type of construct classification rules, the two the the standard close neighborhood and fuzzy membership function The classifier is used in conjunction with the extraction mining covering information. 5) classification accuracy and kappa coefficient as an indicator to evaluate the accuracy of the classification of the two classification methods. Visual interpretation results of the field verification of the reference data to verify the accuracy of location and polygons area polygon object-oriented classification results. Through the study found, based on the classification of the object-oriented compared to significantly higher classification accuracy of maximum likelihood classification, the visual effect. Mining remote sensing survey and monitoring process, the method can play the role of auxiliary classification of man-machine interactive interpretation method to alleviate the pressure of the field site verification, improve indoor interpretation accuracy and timely help to the understanding of the mine development area of ??information.

Related Dissertations

CLC: > Astronomy,Earth Sciences > Geology > Geology, mineral prospecting and exploration > Remote sensing exploration
© 2012 www.DissertationTopic.Net Mobile