Research On Information Extraction Of Urban Built-up Area By Using Medium And High Resolution Intages

With the accelerated process of urbanization,the number and area of urban built-up areas have increased dramatically.Urban built-up area has a great influence on the development of urban economy and the urban ecological environment.By using remote sensing image to extract the information of urban built-up area,we can quickly understand the distribution range and distribution characteristics of urban built-up area,and calculate the area of land occupied by urban expansion and built-up area.For high resolution image of high cost,time series of short and medium resolution image width wider,longer time series,but not for land cover types for accurate extraction,especially mixed pixel resulting in city or adjacent land cover types boundary fuzzy and extraction area with greater error.This paper combined with the merits of the two kinds of images and based on mixed pixel in image relationship model,using medium resolution images of urban built-up area of accurate information extraction,to provide a reliable source of data for urban land use planning.The following main conclusions are obtained through the middle and high resolution image model:(1)By high resolution images of urban built area accurate information extraction,using a random forest classifier and oriented object classification to extract information from the image,using field sampling data classification image to evaluate the accuracy of,through subjective and objective evaluation on comparison,we conclude that the use of high resolution images in urban built-up area extraction using oriented object method is more appropriate.(2)The degradation of high resolution image and the establishment of a linear model.Scale degradation refers to the creation of grid on grid area statistics,the calculation of the new attribute to grid attribute connection,to create a grid of attribute values.Based on the knowledge of mathematical statistics and a large number of experiments,a multivariate linear model is established.(3)To verify the simulation of a linear model.Validation includes partial validation and global validation,partial validation to verify the model's in sample validation R2 for 0.6548 and RMSE is 0.05447 and validate the equation slope 1.68,and the isocon line deviation is far,after the model of system error correction to ensure the accuracy of the simulation results.Global inversion verify that the relative error of the total error of 10%-50%verification 63%.The experimental results show that the linear model is feasible.