Targets Tracking Technology Based On Computer Stereo Vision

Based on computer stereo vision,3-D restruction technology is an important branch of the disciplines of computer vision,and widely used in measuring and drawing tasks of industrial pieces,relics,terrain,buildings and other objects.Compared with the traditional manual measurement,this technology has its own characteristics,such as non-contact,high precision,high efficiency,complex model adaptive.During the measuring process,how to position the camera is a big problem in this area.A set of algorithms are proposed in this thesis by tracking camera.The main work of this thesis consists of the following three aspects:First,an effective algorithm framework(algorithm 3.1)is proposed which cansemi-automatically find the centric coordinates of a target in an image.This algorithm composed of a set of classic algorithms,including Auto Levels algorithm for image,Gaussian smoothing filter algorithm,gradient algorithm,Canny algorithm,the HSCP algorithms,ellipse fitting algorithm and ellipse parameters backcalculation.Second,a rapid global target searching algorithm is proposed(algorithm 4.1).The algorithm searchesall of the targets in the first image,then predicting the possible positions of all targets by using the moving information provided by continuous images.After that,the predicted images are corrected based on the current images,and finally we can get the confining box of the target ellipse(algorithm 4.3),which can provide the initial data for target calculation.Finally,a camera tracking method based on solo camera is proposed.This method uses only one camera,and introduces the fixed target.Thereafter,the camera can be tracked and the coordinates can be found based on the constraint distance between them.Numerical experiments show that,the first algorithm can find the coordinates of the targets in high efficiency and accuracy.In this thesis,the second algorithms provide the possibility of automatic tracking the targets.The third algorithms greatly simplifies the traditional tracking method,and achieve good results in the performance under the condition of a slight loss of accuracy.