Design And Realization Of Parts Geometry Precision Measuring System

With the development of production and technology, the improvement of machining accuracy and the change of process i n g object, people's demand on the measurement accuracy, efficiency and automation is higher and higher. The most key problem of part geometry measurement system is to chieve fast and accura t e measurement of part shape size is. In this article, a common parts size high precision automatic online measurement system was develop e d using key computer vision technologies, such as subpixel edge detection, image registration and camera calibration. This system aims at the lack of the traditional measurement in small, soft, specific shape parts. The main work is as follows:1. Research on subpixel edge detection algorithm. On the one hand, the edge location technique based on subpixel was studied, on the other hand, experiment to compare with fitting, gray moment and spatia l moment subpixel detection algorithm's accuracy and efficiency was designed. In the experiment, to choose a subpixel edge detection and localizat ion algorithm, which can satisfy the real time measurem e n t accuracy, speed, stability at the same time, the main facto r s influencing the accuracy were analyzed quantitatively.2. Design of a high- precision parts image registration algorit hm. A rapid and high-precision parts image registration algorithm was designed based on the center of mass and the minimum circumscrib e d rectangle. The registration can be divided into initial registration and accurate registration. Initial registration improves the registrat i o n speed by using minimum circumscribed rectangle parts and ima ge centroid that cut registration parameters. Accurate registrat i o n improves registration accuracy by regarding mutual information as similarity criteria. The experimental results show that the algori t h m can reach the demands on time efficiency and matching precision of the batch measurement system.3. Development of parts geometry precision measurement syste m. First of all, the struct ure of image measure ment system hardware and software was analyzed. Then, basic subpixel edge detection and ima ge registration algorithms were chosen. At last, the parts geomet r i c dimension measuring system which should possess founctions of measuring a single part, measuring batch and statistic analysis was designed and developed applying the technique above. Testing result s show that the system repeated measurement precision can reach 5 ?m.