Three Dimensional Accurate Positioning And Application Based On Computer Vision

With the development and progress of Internet and computer technology,computer vision in various fields has been widely used,various fields such as aerospace,medical equipment field,text document analysis,intelligent robots,human-computer intelligent interaction and visual traffic surveillance and so on,so computer vision has very broad application prospects.However,camera calibration in computer vision research is first needs to solve the goal,the camera calibration task is determined camera parameters,the camera parameters including the main point and focal length of the camera related to the internal structure of the internal parameters and also includes a different image to the camera motion parameters(the camera external parameters).Accurate calibration results have important role in 3D reconstruction,so the simple,fast and high precision camera calibration method has become an important direction of research on camera calibration.The for camera calibration in the process of image processing of image preprocessing technology,to raise the camera calibration accuracy.The classical Zhang Zhengyou camera calibration method of feature point positioning accuracy is improved,mainly joined based on SIFT operator corner detection and the least square fitting,according to the principle of least squares fitting,an attempt was made to the of local curve and curve fitting by global fitting blend together to corner detection scheme,obtained the good effect.On the basis of summarizing the traditional evaluation method,the method of weighted average error is put forward to evaluate the result of the camera calibration.Experiments show that the method has strong robustness and can be obtained in the interference(such as camera imaging due to illumination noise influence,etc.)more strongly,the corner features are extracted,and can of camera parameters with high precision.One of the important purposes of visual orientation is to determine the relationship between the real objects and the orientation of the camera in real life.In this paper,a new method of monocular vision based on computer vision is presented.The method is to target quadrilateral each vertex processing,according to the feature points between the vector addition of the parallelogram law to calculate the camera coordinates of the object pose closed form solution,obtained by running the matlab program specific experimental results.By experiment show that the method has high location accuracy,calculation is convenient,and can realize the visualization of target positioning,and successfully applied in 3D reconstruction,which shows that the theory and the application research results has certain practical value.