Extraction of Urban Green Area Based on Object-oriented

At present , with the rapid development of remote sensing technology especially the improvement of remote sensing image processing , many cities of in china and overseas have applied remote sensing to greenland information extraction ,in order to find out green area cover dynamicand optimize the spatial structure of green space.This paper takes high-resolution remote sensing image as the data source and extracts the information of the urban green area according to the method of object-oriented method . The texture feature is extracted by making use of the grey co-occurance level matrix and wavelet transform .The extraction result is compared to the traditional method to evaluate the accuracy . Study the meaning of the extraction of urban green area based on object-oriented method ,and get an effective extraction method of the urban green area . 1. FOREWORD With the rapid development of remote sensing technology especially the improvement of remote sensing image processing , remote sensing has been applied in the various social fields more and more widely increasing .The urban green space is an important part of the ecology system , it is the result that many factors take part in ,for example, nature and humanities and so on , and it is the sign of the environment of city and the living standard of the people .In the recent years , many cities of in china and overseas have applied remote sensing to greenland information extraction ,in order to find out green area cover dynamicand optimize the spatial structure of green space. It can increase the city’s potential of sustainable development and realize the planning of the green space. Traditionally the information extraction of the remote sensing image mainly has two methods: the pixel-based approach and the manual interpretation approach .The extraction of the urban green space is based on the pixel and carries the classification and extraction according to the spectrum features. Since this technology a pixel as the unit , its result of the extraction is dispersed , so it influences the extracted accuracy seriously . Along with the development of the remote sensing technology , more and more remote sensing images with the high-resolution have been applied to the extraction of the green space in the city. The high-resolution image implies abundant texture features and spatial structure information ,so the high-resolution image processing based on pixel will produce “salt” phenomenon .The accuracy of the extraction result is lower because of the noise . In order to achieve the information extraction of the high-resolution image , rational use the bundant information of the high-resolution image , according to the features of high-resolution remote sensing images, the object-oriented information extraction methods have emerged. 2. THEORY The paper extracts the information of the urban green area according to the method of object-oriented method . Firstly it carries on the pre-processing to the remote sensing image , Secondly the integrate criterion of the smallest heterogeneity and the scale parameter are applied to the segmentation , it divides the image into many homogeneous regions. Thirdly extracts the texture features by making use of the grey co-occurance level matrix and wavelet transform . Then target at the segmentated image, make use of the object-oriented method to combine the textures extracted from grey co-occurance matrix and wavelet transform to classify and extract green area . 2.1 Image Segmentation The key of the method based on object-oriented is image segmentation ,the important step is how to choose the appropriate scale of the image segmentation and extract the target . Multi-scale image segmentation is that the same image is segmented several times with the different scales. So the ground object and structure information can be described by the s segmentation results of the different scales .Multi-scale segmentation is adopted the calculation of region growing of the smallest heterogeneity .At the process of the segmentation consiers not only the spectral characteristics but also the spatial features and the shape features(Yiqun Xiong and Jianping Wu,2006) . Spectrum factor and the shape factor can be determined at the segmenting ,and the shape factor contains the compactness heterogeneity and the smoothness heterogeneity(Zhenyong Zhang and Wang Ping,2007) . The calculation formula of the heterogeneity f can be expressed(Xiaofang Sun and Lu Jian,2006):