Study On The Methods For Extracting City Green Land Imformation Based On The High-resolution Remote Sensing Image

Green land is the purifier of the city, which has an ecological function and animportant social function. As an important part of the urban ecosystem, green land play apivotal role in the adjustment process of the urban environment. With the development ofthe remote sensing technology, it has been extensively used in all fields of society. Theinformation of green land can be obtained rapidly, objectively and accurately by usingremote sensing techniques. In an addition, the basic data of the distribution and area of thecity green land can be obtained rapidly with the remote sensing techniques. So that it canprovide better planning basises and management decisions for the development of the cityand has an important practical significance for improving the sustainable development ofthe city.In this paper, the SPOT5remote sensing image of Qingdao City was selected as a datasource to be studied for the green land extraction. First of all, a series of pretreatment suchas integration, calibration, cut and spectral enhancement were done for the images. Then,the green land information of the study area was been classified and extracted with threeclassification methods including ISODATA classification, maximum likelihoodclassification and object-oriented classification, and the classification results obtained fromthe three methods were compared. The results showed that the classification accuracy forthe ISODATA method, the maximum likelihood method and the object-oriented methodwere61.3%,80.0%and88.7%, respectively. As seen from the classification effect andaccuracy, the result from the object-oriented method was better than the maximumlikelihood classification method, which is better than the ISODATA classification method.Finally, the green land information of the SPOT5remote sensing images of the study areain different time periods was extracted using the objected-oriented classification with thehigher accuracy. Additionally, the changes of the green land information from two timeperiods were visually monitored with calculating the statistical area.In this paper, when the city green land information was extracted with theobject-oriented method, the rule-based fuzzy classification method was used, for which the membership degree was used to express the probability that the image object belonged tothe some class. Therefore, it can avoid that using “yes” or “no” to express the satisfactionfor the described features, which is more accordant with the expression of the uncertainty.