Information Extraction And Landscape Pattern Analysis Of City Park Green Sace Sstem Based On High Resolution Remote Sensing Images

Urban green space, as "the lungs of the city, has an irreplaceable role in the economic,ecological and other aspects. Green parks as an important part of urban green space system,and its role should not be ignored. To achieve the planning and construction of urban parksand green space, we need to be efficient, fast, high-precision extraction of urban green spaceand information. Remote sensing technology, geographic information system technology andglobal positioning system technology development, as well as high-resolution satelliteimagery provides an avenue for high-precision extraction of urban green space andinformation.April2005, SPOT-5image as the main source of information for the study area toYanzhou shaoling Park and we take the pretreatments of the original data of remote sensingimage SPOT-5atmosphere correction, radiation correction, geometric correction, cutting byusing EARDAS, ENVI, ARCGIS and related software. On the basis of the comparative studyof image fusion technology, the fusion of SPOT-5high-resolution panchromatic image andlow resolution multi-spectral band image has been come true. And then using a variety ofmethods to classify and extract the green vegetation of the the Yanzhou shaoling Park. Thenby the effect of contrast and precision of analysis,we adopt the most effective support vectormachine based on spectral and texture features of a green landscape extraction andclassification steps,then we get a feature type in of Yanzhou shaoling Park map and greentypes of information map.On this basis, we calculate the number of landscape index byFragstats, Excel and other software patches on the type of landscape three levels, and the weanalyze the park’s landscape spatial patterns of the of Yanzhou shaoling Park landscapepattern of the problems and provide a scientific basis for the Yanzhou park green spacestructure and the pattern of perfect.The main conclusions of this paper:（1） Based on SPOT-5image fusion technology, this paper analyzes a variety of fusionand poses fusion evaluation. Experiment uses Mutiplieative transformation method, Broveytransform method, the PCA transformation method and Pan sharpening transform method tofuse SPOT-5high-resolution panchromatic image and low resolution multi-spectral band image. Through evaluating fusion image qualitatively and quantitatively, it calculates thecharacteristics of the image mean, variance, information entropy, average gradient and thecorrelation coefficient. It is found that Pan sharpening methods takes advantage of thehigh-resolution image fusion by comparison.（2）This paper extracted the green vegetation by using supervised classification andunsupervised classification based on the vegetation index, the texture classification andsupport vector machine based on the multi-feature on the Yanzhou shaoling Park2005high-resolution SPOT-5remote sensing image. By contrasted the results and the accuracy, theresults showed that, the method of support vector machine based on spectral and texturefeatures was one of the most effective green vegetation classification and the overallclassification accuracy was up to91.67%.At the same time, by this method, we got the featuretype map and green types of information map of the study area.（3）Using the principles of landscape ecology and field survey results and on the baseof using high-resolution remote sensing images,a variety of technical approach and theclassification of the landscape elements of the of Yanzhou shaoling Park, this paper calculatethe area and the density index, shape index, adjacent index, diversity index, contagion indexlandscapes index, respectively, in patches, the type of landscape three levels to analyze thepark landscape spatial patterns by using the FRAGSTATS3.3GIS graphical data. The resultsshowed that of Yanzhou shaoling Park landscape elements plaque layout is basically rational,landscape structure is more stable and green patch of the different types of distributionpatterns, shapes are concentrated and the dispersed phase mixing model which is in line withthe requirements of the ideal landscape pattern model.The green landscape belongs to the coreecological plaque, and the environmental quality of the whole park has an important role.