Research On Video Moving Object Detection And Positioning Methods

Video image moving target detection has always been an important development direction in the field of computer vision,and it has been widely used in maritime closed circuit television monitoring system,intelligent transportation and other fields.Commonly used detection algorithms are:optical flow method,median algorithm,mean algorithm,interframe difference method.Because slow moving object is similar to background,so the traditional object detection algorithms are not sensitive for slow moving object,and prone to miss object.In view of this problem,the characteristics of the traditional detection algorithm are analyzed,the appropriate image preprocessing technology is selected,and the traditional interframe difference method is improved in this thesis.By analyzing the correlation of the target between current and previous,the dichotomy convergence algorithm is proposed to approximate the optimal interframe space.The accuracy of the moving object detection is also dependent on the accuracy of the background.In this thesis,a background iterative algorithm based on time and space is proposed.The adaptive weighting factor and the updating coefficient are used to correct the background,and then the moving target is obtained by the difference method.Finally,the image is segmented by fast local threshold method to obtain the binary image.Visual positioning algorithm has become a research hotspot in the field of intelligent monitoring because of its high application value and good development prospect.In this thesis,the imaging model is established,the corresponding projection matrix is deduced,and the camera parameters are obtained by the three-point calibration method to realize the monocular camera positioning.Experiments show that the improved interframe difference method and the object detection method based on time and space have good detection results for slow moving object,and the detection method based on time and space has better robustness and superiority.The visual positioning algorithm is simple and the error is small.The experiment proves the effectiveness of the method.