Design And Implementation Of A Monocular Flying Target Pose Measurement System

Target pose measurement not only provide theoretical basis for the judgment and recognition, but also provide effective measurements for target testing, how to make an accurate and effective target pose measurement scheme has been a problem in the field of computer vision and photogrammetry. According to the number of the photographic equipment, target pose measurement method can be divided into two categories: multi-eye pose measurement and monocular pose measurement; according to the different algorithm, it also can be divided into two categories: pose measurement based on the study and pose measurement based on model. When multiple photographic equipment are used to shoot for flying target testing, because of the limitation of shooting conditions, only one device can track flying target while other equipment lose the flying target, aiming at this situation, it will make a further research on monocular flying pose measurement issues based on model.It is the main idea to use the relationship between the characteristic points and the control points to measure the target pose. But due to the low quality of the reality picture, it cannot build accurate contour point correspondence to solve the problem of measuring the pose parameters generally. So this paper uses a new method for monocular flying target pose measurement based on model feature matching. Firstly, it use of 3 d modeling tool to import flying target model and use Open GL to simulate flying target motion state. Then the image segmentation algorithm is used to separate the target in the real image and the simulated image from the background. And the contour features are extracted to be used in the next step. With it, the GPFI algorithm can solve the relationship between real target contour and simulated target contour point set, which can effectively avoid the correspondence problem of pose parameters.This paper introduces the pose measurement coordinate system transformation and analyzes the essence of target pose measurement. According to the imaging principle of Open GL, it proves the correctness using Open GL simulation flight target solution. At the end of the thesis, the analysis of the present method and the method of different is presented. The accuracy of the Open GL simulation is demonstrated by the Open GL imaging principle, and the difference between the present method and this method is also showed.By analyzing the shortcomings of conventional edge detection operators, it proposes that firstly use of optical flow combined with Cut Grab image segmentation method to segment the target image, and then use the conventional operator to locate the contour points. Finally, according to the principle of the gray sequence in the normal direction of the contour points, the sub pixel contour extraction method based on Gauss fitting is used to precisely locate the position of the contour line of the real image and the simulation image.Based on the precise target contour, it describes the survey adjustment principle and error equation of the GPFI method. And with the above theory simultaneous equations between real image contour and simulation image contour are founded.Then it uses the improved objective function to assist the GPFI method to determine the iteration so that it can get the target pose parameters by the calculated results of coordinate transformation.At last, the method of this paper is simulated by using the simulation data. The validity and correctness of this method are verified by analyzing the experimental results.