Font Size: a A A

Application Of Texture Information And CART Decision Tree Technology For Image Classification Using Remote Sensing Data In Wenyuan

Posted on:2009-02-22Degree:MasterType:Thesis
Country:ChinaCandidate:Y TanFull Text:PDF
GTID:2143360245456578Subject:Forest management
Abstract/Summary:PDF Full Text Request
A new Earth observation system, which is formed by remote sensing system, global positioning system and geographic information system, provide new scientific methods and technical means to the forest science research. The social and economic benefits are self-evident.Traditional pixel-based classification method depends only on the spatial information ,which leads to much declassification .Especially ,when it comes to the classification of high resolution image ,its accuracy is far from productive demand .Plenty of experiments indicate that the problem of forestry land type identification can not be solved depending only on the spatial spectrum information .Besides ,the remote sensing imagery contains lots of detail information of texture .We can make good use of texture and spatial information during the classification process .Choice of a classification algorithm is generally based on a number of factors ,among which are availability of software ,ease of use ,and performance ,measured here by overall classification accuracy . In the past few years ,the use of decision tree to classify remotely sensed data has increased .In this paper ,we use decision tree classification algorithm ,with texture information ,on the Landsat TM / ETM + images of Wengyuan area .Proponents of the method claim that it has a number of advantages over the maximum likelihood algorithm .Decision trees produced consistently higher classification accuracies than the maximum likelihood algorithm .The texture information also improves the accuracies .Decision trees have the advantage of being nonparametric and therefore make no assumptions regarding the distribution of input data .
Keywords/Search Tags:Forestry Land Categories, Texture Information, Decision Tree Classification, Remote sensing, TM Data
PDF Full Text Request
Related items