Research On The Application Of Object - Oriented High - Resolution Remote Sensing Image Classification In Satellite Law Enforcement

At Present, the application of high resolution remote sensing image is very wide,it is also unceasingly thorough to the classification research of information extraction from high resolution remote sensing image,in order to meet the needs of constantly growing. The high resolution remote sensing images compared with the low resolution remote sensing images, it has the more information of characteristic, the traditional classification methods based on pixels on the classification of high resolution remote sensing image has some defects, it can’t fully utilize and analysis multi-feature of image information, and should reduce the classification accuracy, no longer meet the classification requirements. According to the requirements of satellite image and the characteristic of high resolution remote sensing image classification, the paper proposes to use the object-oriented method to classify and extract the high-resolution remotely sensed data.Taken a SPOT 5 image of certain characteristics information as the experimental data, According to the image information by visual interpretation, select extract building of image information. The experiments with Ecognition and ENVI software as the platform,use the object-oriented classification of experimental data. In the classification processing,according to the characteristics of the objects as first steep, using multi-scale segmentation, constructing the classification system, and using the fuzzy classification method to extract buildings. Secondly, the experimental data using the traditional pixel oriented classification methods (maximum likelihood classification, minimum distance classification and Mahalanobis distance method) compared the classification results with object oriented remote sensing image,and analyzed the results of two classification methods, due to the multi-scale segmentation image, using object oriented classification method extract the remote sensing image information, the building features not distortion and improve the accuracy of the classification, the classification results can effectively avoid the "salt and pepper noise", the results of classification have consistency in shape and property.