Monnocular Vision Based On Mobile Robot Localization Method Of Target

Location has become an urgent problem to be solved in many areas, the accurate pose estimation is becoming important to the generation of automatic map, path planning, target detection and tracking. Because the photographic images contain a large number of environmental information, it is paid a lot of attention, thus the location technology based on monocular vision has become a research focus.There are many methods of the measurement location based on monocular vision, such as the geometric similarity measurement, structured light measurement, geometric optics measurement, aided measurement and laser-assisted distance measurement. If don't consider other aids, only using the monocular camera to achieve the location measurement, there are mainly two methods, which are the geometrical similarity measurement and the geometrical optics measurement.The geometric similarity measurement requires that the distance between the camera and the measurement plane is fixed, and also the relation between the camera and the measurement plane is vertical. But usually there is a pitching angle between the mobile robot based on the monocular vision and the target, so this paper put its focus on the geometrical optics measurement introduction and how to use it.The mobile robot based on the monocular vision measurement location method is mainly discussed in this paper, its goal is to achieve the measurement of the 3D target object position and the attitude real-timely. The main research contents are: Through the FPGA control video processor to collect the image data; through the correspond image processing technology to get the dimensional characteristic information of the target object in the image space; based on whether the camera and the target is vertical, to calibrate the camera parameters; using a new and simple and feasible location algorithm which based on the target perspective projection imaging model, to get the space pose of the target.