Design And Implementation Of Plastic Nozzle Defect Inspection System Based On Machine Vision

Plastic nozzle is widely used in modern industrial production. However, due to variety of reasons from manufacturing factor and production technology, it is easy to produce a variety of bubbles and other defects, which seriously impact on product quality and its application prospects. On the other hand, artificial detection bubble defect on plastic nozzle process scene is almost impossible, which will not only seriously affect the workers health, but also human visual ability cannot effectively match the requirements of modern production speed and accuracy.Machine vision technology has the characteristics of full automatic, high speed, high stability, high precision, high resolution, and what is more, very small interference to the measured target. It has very widely application prospect in defect detection field, and is very suitable for plastic nozzle production process requirements.This thesis has designed and demonstrated the nozzle defect detection system scheme from hardware and software the two aspects. Firstly, according to the process scene and technical requirements, by detailed analysis to existing plastic nozzle production and detection process characteristics, we find the difficult in the plastic nozzle bubble detection process. At the same time, this paper also executive the function analysis and equipment selection demonstration to the hardware component model of machine vision detection system. In the design of visual software system, first of all, by use of the image processing technology such as the high lift filter and mathematical morphology, we carry on the pretreatment to the image, to improve the performance of machine vision, finally an improved edge enhancement filtering method has been proposed, successfully realize the extraction of various bubbles in a plastic nozzle. The analysis to experimental results show: compared with traditional based on gray morphology and region growing method, this method has obvious superiority to extract various types of bubbles in plastic nozzle.The project designed and constructed a plastic nozzle defect detection system based on machine vision principle, according to the physical properties of plastic nozzle and the survey results to the actual production process scene. Its detection rate achieves 200/min, the detection error rate below 0.3%, the missing rate is below 0.1%, the minimum recognition area is 0.5mm2. The system has been applied in the plastic nozzle production site of enterprises, which has proved that this technology can quickly and effectively implement of different types of plastic nozzle bubble defect detection, achieve a relatively good performance.