Study On The Extraction Of Urban Typical Targets And The Functional Classification Of Urban Blocks From High Resolution Remote Sensing Imagery

The industrial revolution has enabled the rapid development of urbanization in the world.However,under the guidance of the urban planning idea of some non-integrated functional partition,the most basic functions of the city are forced to be separated from each other,So that the internal function of the loss of activity,lost for the purpose of human life services.As the basic unit of the city,the block is the basis of the economic,cultural and political activities of the city.Therefore,it is possible to study the urban model by studying the urban block model and find out the problems in the city.However,due to the large amount of information and the wide range of urban blocks,resulting in access to urban neighborhood information needs to spend a lot of manpower and material resources,and long cycle,less efficient.With the rapid development of electronic technolo gy,optical imaging technology,network transmission technology,remote sensing has a multi angle,high spatial resolution,high temporal resolution,high spectral resolution and other advantages of earth observation technology,and can obtain detailed ground objects information and short cycle.Based on the above analysis,the paper proposed a method for the extraction of urban typical targets and the functional classification of urban blocks from high resolution remote sensing imagery.The results of the study include the following aspects:(1)A new object-oriented method of shadow extraction from high resolution remote sensing imagery is proposed.First,the mean shift algorithm is used to cluster the features to remove noise.Second,shadow detection was performed using the Shadow Detection Index proposed in this paper.Finally,the shadow region is extracted by threshold segmentation.The experiments were performed using two remote sensing images of differe nt scenes and different sensors.The experimental results show that the method can extract the shadow information accurately and effectively,and can remove the influence of non-shadow objects such as water body and blue objects.In addition,the use of object-oriented thinking can effectively remove the influence of noise and improve the detection accuracy.(2)We present a method for automatic extraction of main roads from high resolution remote sensing imagery in this paper.The proposed method is composed of three parts.First,the algorithm,Line Segment Detector,is used to extract the line segments from the images.Second,perceptual organization is realized to cluster the line segments.Finally,the road information is obtained through length constraint.The experiments were performed using two remote sensing images of differe nt scenes and different sensors.The experimental results suggest that the completeness,correctness and quality of our proposed method on the two images are all larger than 96%.(3)The paper proposed a method for functio nal classification of urban blocks from high resolution remote sensing imagery.First,the characteristics of buildings,vegetation,water body and shadow are obtained.Second,the city is divided into independent block image units by combining the road network information,and the pixel-oriented approach is transformed into the processing method for block objects by calculating the mean value of each feature image of each block.Finally,the classification of urban neighborhood functions is realized by LIBSVM classifier.The experimental results show that this method can be used to classify the eight different functional blocks,such as the village,modern residential,commercial,etc.,and the accuracy is more than 84%.