A high accuracy position method based on computer vision and error analysis

The study of high accuracy position system is becoming the hotspot in the field of autocontrol. And positioning is one of the most researched tasks in vision system. So we. decide to solve the object locating by using the image processing method. This paper describes a new method of high accuracy positioning method through vision system. In the proposed method, an edge-detection filter is designed for a certain running condition. Here, the filter contains two mainly parts: one is image-processing module, this module is to implement edge detection, it contains of multi-level threshold self-adapting segmentation, edge-detection and edge filter; the other one is object-locating module, it is to point out the location of each object in high accurate, and it is made up of medium-filtering and curve-fitting. This paper gives some analysis error for the method to prove the feasibility of vision in position detecting. Finally, to verify the availability of the method, an example of positioning worktable, which is using the proposed method, is given at the end of the paper. Results show that the method can accurately detect the position of measured object and identify object attitude.