Vision System With Humanoid Eyes And Neck And Target Positioning Algorithm

For humankind, acquiring information by vision from circumstance is one of the most important sensing methods. In the same way, machine vision is one of the most important sensing methods for robots to get information from circumstance. Machine vision system is a synthesis of hardware and software, which can perform certain vision functions. After researchers'several decades'effort, machine vision develop rapidly, and it goes into practical application from laboratory. But to be objectively speaking, machine vision is still in immature stage. The mainly purpose of this paper is to research the application of machine vision in target location based on bionics of human vision.The structure of the vision system with humanoid eyes and neck in the research is introduced at the first, and then an overall frame of the research is designed, according to the target location task and the experiment condition. The system is divided into many basic blocks whose mainly functions and realized approaches are discussed.In the aspect of machine structure of the vision system, vision system with humanoid eyes and neck which can imitate human manner of moving was designed basing on bionics. This vision device has eight degrees of freedoms. In the aspect of hardware control of the vision system, four degrees of freedoms of the vision device was completed to achieve that degrees of freedoms of vision device can move promptly and in phase in the experiment of target location. In the aspect of upper computer software of the vision system, functions such as the communication between upper computer software and hardware, the show of the picture and the transfer of the arithmetic were accomplished, so that, the operation and control of the vision device will be visible and convenient.For locating the distinguished target on the center of the camera, target positioning algorithm and positioning error compensation algorithm were proposed. Target positioning algorithm can transfer pixel coordination of the target into turning angles of degrees of freedoms by way of transformation matrix and parameter calculation basing on camera calibration technical. The machine structure of the vision system is of multiple degrees of freedoms and the control system is of open loop system. In order to acquire precision request of experiment, positioning error compensation algorithm improve the precision of target location according to the attitude information of vision device. The experimental results show that the design of vision device and the positioning algorithms have acquired the request of the static target location and dynamic target location experiment, realize the location task of distinguished target.