Study On The Image Processing Methods Of The Online Detection System Of Conical Thread

Since the quality of tubing threads'connection has great effect on normal oil production,at the same time, the thread measuring method is still staying in manual method, which has high error possibility and low precision. Now we need to study a more effective thread measuring system suitable for stable production.In this thesis, we designed a system for auto-measuring parameters of outside screw based on non-contact measuring and image processing techniques. The measuring system researched is on the ground of optics, mechanics, electronics and mathematics.The main content of this paper includes: the overall design of the measuring system, image acquisition and preprocessing, screw image segmentation and edge detection, calculate method of screw parameters and facial look judging .With the experiment analysis, we fount that the SUSAN operator is effective to detect the edge of screw image, and mathematical morphology and sub pixel are also applied to process the edge. Least Squares procedure is employed to fit the external screw profile line. So we can calculate geometrical parameters of the screw with a high accuracy by the implemented system. At last, according to the computed parameters and the facial look characteristic of the screw, the system can carry on judging to discard automatically. The experiment result gives satisfaction.