Font Size: a A A

The Research On Feature Extraction The Wetland In Poyang Lake Area

Posted on:2015-01-30Degree:MasterType:Thesis
Country:ChinaCandidate:L LiFull Text:PDF
GTID:2180330467988472Subject:Cartography and Geographic Information System
Abstract/Summary:PDF Full Text Request
With remote sensing image appling more and more widely, using remote sensing imageto extract wetland characteristics and manage the wetland information scientifically is animportant means of protecting wetlands and wetland species. Poyang Lake wetland is thelargest in Asia, it has rich resources, many types, and very representative. Therefore, usingremote sensing technology to extract the characteristics of Poyang Lake wetland has practicalsignificance.This paper firstly introduces the research status about the remote sensing imageclassification and wetland information extraction at home and abroad. Secondly, it makes ageneralization of the general situation of the study areaļ¼Œpreprocesses the image data andcollects the samples. Thirdly,the paper uses the ISODATA, K-Means, parallelepiped,minimum distance method, maximum likelihood method, decision tree algorithm and BPneural network algorithm to extracte wetland characteristics of Duchang county, gets theresults and evaluates their classification effects. Then, making study of the integration of thedecision tree and the BP neural network algorithm, it puts forward a kind of integrationalgorithm, used to extract the wetland characteristics of Duchang county, gets the results andevaluates its classification effect. Finally, it analyzes the classification accuracy of all themethod used in this paper comprehensively, and uses the integration algorithm to realize thewetland feature extraction of Poyang Lake Area.From the results, the overall classification accuracy of ISODATA method, K-Meansmethod, parallelepiped taxonomy, minimum distance method, maximum likelihood method,decision tree algorithm, BP neural network algorithm and the integration algorithmrespectively is75.58%,82.16%,75.46%,77.63%,82.69%,77.63%,89.96%and92.93%. Theresults show that the integration algorithm based on decision tree and BP neural network issignificantly higher than the other methods in wetland feature extraction, providing a newway for the wetland informationĀ·s extraction.
Keywords/Search Tags:Poyang Lake, Wetland, Feature extraction, Algorithm, Integration
PDF Full Text Request
Related items