Based On The Detection And Damage Identification Of High-resolution Images Of The Airport Area

The technology of object automatic recognition on remote-sensing image is avery active research field in pattern recognition and image processing these years.Along with the widely use of military object recognition, the research of the automaticrecognition of battle damage of airports has great meaning to the military. This papermakes Some researches and progress on the object recognition based on some currentalgorithms, and achieves some results.This paper focuses on the feature extraction of linear object of airports anddesigns a airport recognition system through straight linear shape based on the result ofshape analysis. A method of line detection based on improved arithmetic ofmultiresolution Hough transform is used to detect the runway. The line detectionmethod can make the detection faster and reduce the storage capacity obviously. Thenaccording to the knowledge about the structural information and intensity feature ofairport, a series of rules are to determine the airport in the remote sensing images.In this thesis, we also study recognition of battle damage in the valid airportregion and introduce two methods to detect the battle damage of airports after strikes.The first one is unsupervised change detection algorithm which compares theairport images taken before and after strikes, and gets the battle damage informationaccording to the difference of the images. First we divide the images into small parts ofthe same size, and detect the change region based on image matching. And then amethod based on Bayes decision for the minimum error is used for accurate battledamage detection.The other one is crater modeling recognition. The process is composed of threemodules, the feature matching of craters, region growing of craters, and edgedescription of craters which uses curve fitting model based on Least squares method.Experimental results demonstrate that the methods described in this thesis areeffective for the automatic airport detection and battle damage recognition.