| Object recognition is one of the important parts of computer vision, and has been widely applied in the fields such as military affairs and automatization. The recognition of airplane is one of the applications of object recognition. There are many methods for the recognition of the friend-or-foe airplane, but it is just in the groping stage to recognize airplane mode. Correct recognition is key in information processing in the late period of command automatization system and also important in civil aviation filed.In this paper, we study the method of recognizing airplane in its 2-D image. The projective transform occurs in the 2-D imaging process, therefore in order to recognize airplane object, we have to find projective-invariant shape descriptor of object. It is the core of this paper. Firstly in the stage of image segmentation, on the basis of comparing three edge detector in common use, we propose a new method of edge detection, which is combining Robert detector and contour tracing. After getting chain code in the stage of image segmentation, we introduce.the concept of chain code sum, and propose the method of detecting projective-invariant key points on the contour of airplane. On the stage of describing the shape of airplane and extracting feature, we derive five-coplanar-point invariant from the basic projective invariant-cross ratio, and resolve the problem of stabilization of five-coplanar-point when there is three points are nearly collinear. And then we study the stabilization of five-coplanar-point in the case of noise, and work out the criterion utilized to select a stable invariant. In the stage of recognition, we utilize the model-based recognition tactics. We utilize hash table to build model-based database and give the algorithm of building and searching model-based database. Finally we testify the validity of the variant proposed in this paper.This thesis is supported by Graduation Seed Fund of Northwestern Polytechnical University(Z20030060)...
|
PDF Full Text Request