The Real-time Auxiliary Materials Inspection Technology Based On Machine Vision

Aiming at the requirement of rapid detection of mobile phone accessory assembly defects on the production line,this paper designs a mobile phone accessory assembly defect detection software based on video information.The software realizes the automatic capture of mobile phone images on the production line and the local feature detection of various accessories,including: color distribution based discrimination method,edge-based discrimination method for transparent accessories or edge for large-area sandblasted logo defects.Projection decision method,local feature extraction and matching based decision method for character string printing;sample feature collection and statistics function is designed to help operators reasonably set decision parameters;image quality evaluation and peripheral device control functions are designed to adjust camera Parameters,lighting brightness and control cylinders push the defective products away from the conveyor belt;design database management function,which can add,edit,delete,etc.data of production logs,defect pictures,test templates,etc.of different manufacturers,different products and different time..The paper focuses on two technical difficulties encountered in the design process.A fast target acquisition algorithm in image sequence for target detection on production line is proposed.It is characterized by using target positioning state to obtain effective information.When the target enters the field of view,the historical multi-frame image is used to estimate the target moving speed and direction.The algorithm can adapt to changes in the speed and direction of the production line,and demonstrates better robustness and real-time response capability.When the target is completely present in the field of view,the previous target is captured using the previous target and the target moving speed is predicted.The position in the current image,and then whether the current target is a new target,solves the problem of capturing the same target multiple times in the field of view,and achieves rapid capture of multiple targets in the production line video sequence.Another key aspect of mobile phone accessory testing is the accurate matching of the captured image of the tested phone with the standard image of the qualified product.Aiming at the problem of image registration deviation in actual test,an image registration algorithm based on two local feature points extraction is proposed.By increasing the mask area,the extracted feature points are evenly distributed in the image.The experimental results show that the proposed algorithm can significantly improve the image registration accuracy,and has good robustness.It can also obtain good registration accuracy under the conditions of strong illumination,such as uneven illumination,blurred photographs and significant background clutter.The overall software of the mobile phone accessory inspection system designed in this paper has been developed and delivered for use,to meet the real-time detection requirements of the transmission line,and the detection accuracy rate has reached the expected technical indicators.