Research And Realization On The Recognition Of Airport And Bridge In Remote Sensing Image

With the development of remote sensing technology, many high-quality remote sensing images could be obtained easily. More and more researchers focus on the automatic detection and recognition of objects in remote sensing image. Airport and bridge, which are the typical targets, become important objects in military research. Lots of scholars propose many effective recognation methods. Unfortunately, for the complexity of remote sensing image, there has been no effective and adaptive ones to our best acknowledge yet.On the basis of previous studies, this thesis implemented a detection method of mutiple airport based on SVM. Meanwhile. The definitions of the density of edge and the complexity of segment is given and a new detecion method of bridge, without water under it, based on the definitions is proposed.In the detection method of mutiple airports based on SVM, firstly, the source image is segmented by WFCM algorithm with one-dimensional histogram and the skeleton of the segmented image is extracted. Secondly, the edges are obtained by Canny operator and straight lines are gotten with Hough transform. Finaly, identify the suspected airport area are distinguished with runway extension and acess roads extraction. Fifteen features of the extracted suspected airport, such as area length, width, acess roads and Zernike moment features, constitute a feature vector, and the area could be justified by SVM based on this set of feature vector.In the detection methord of the bridge wihout water, firstly, the edge is etracted by using Canny operator, and the density of edge for each pixel is calculated, then the image is segmented base on the density of the edge. Secondly, the suspected bridge area is identified by a series of processing including using Hough transofrm to extract straight line and calculating segment complexity. Then ten texture features of the suspected bridge area, such as entropy, energy and releavance, are calculated to constitude a feature vector. Finaly, classify each suspected bridge by using the BPNN on this feature vector.The experimental results show that both the two algorithms can accurately detect the target in remote sensing image. Detection algorithm for airport solve the detection problem for the image which contains more than one runway. And the other algorithm based on the density of edge and the complexity of segment provides a new idea for the decetction of this bridge.