Research On The Classification Method Of Urban Land Based On The QuickBird Image

In recent years,with the rapid development of science and technology,various kinds of resources of the satellite launch,human real-time access to the surface of the various real-time information,the development of remote sensing industry is also growing,hyperspectral,high resolution and other remote sensing images also appears.The high-resolution remote sensing image is still lagging behind,because of the large amount of information,the complexity and the complexity of the texture change,the high-resolution remote sensing image has high speed,high spatial resolution and detailed geographic information.Resolution image data processing and information extraction has also been troubled by many researchers,which also makes high-resolution remote sensing can not be widely used.But there is no doubt,because of the high resolution image has many advantages in the future will become an important direction of development of the remote sensing industry.At the same time with the development of society,the city as the most gathering place of human life,its information acquisition,update has become particularly important,with a variety of high-resolution remote sensing image data,to the urban land use change research provides a Good data sources,especially for urban areas.Because of the high resolution of higher quality feature information extraction,the characteristics of space information is more abundant and obvious,which can greatly improve the classification accuracy and the high resolution remote sensing image information extraction technology has become one of the city land use.This paper takes the QuickBird image as the research data classification method based on random forest,the random forest classifier model was established through experimental design,the optimal parameters of the classifier is established.At the sametime in the experiment on the extraction of the features of rendering images in different segmentation scales are analyzed and compared.The results show that the segmentation scale in 50,the compact degree is 0.3,and the spectral factor is 0.5 under the condition of the highest classification accuracy,the feature information of city classification.And at the same time with support vector machine classification and machine learning over the results of these two methods are analyzed and compared.The results show that the error classification in the RF classification results is the least,and the zero error situation of other objects can be realized when extracting a certain object,and the accuracy of the classification result is greatly improved compared with the other two classification methods.the classification results The best.In this paper,the remote sensing image with the highest resolution and the most effective classification model are combined,to achieve the classification of city land use,to verify the efficiency of machine learning method in remote sensing image classification applications.