摘要：Thread inspection is an important part of modern industrial inspection processes and the key to product quality. Traditional inspection methods, which are mostly of the contact type, are time consuming and expensive. Therefore, this study developed an automatic on-machine, non-contact measurement system for thread profiles by using eye-in-hand machine vision with morphology technology. On the basis of machine vision, this study proposed a method of measuring thread dimensions to implement non-contact, rapid, accurate measurements. The measurement samples were metric threads, and the images of thread profiles were obtained by backlight during the measurement process. Numerous image processing and machine vision methods were developed to analyze the thread images and thread profiles and calculate the depth and pitch of threads. This study also clarifies the steps involved in the selection of the imaging system, hardware settings, image capture, and information processing for the measurement system. The developed system-measured dimensions and 3D microscope-measured target dimensions were compared and validated. Results showed that the maximum difference between the target and measured dimensions was 6 μm (one pixel). The measurement difference for the three workpiece samples was approximately two pixels, thereby validating the feasibility and accuracy of the system and the approaches used in this study.
