Key Technology Research Of Aluminum Surface On-line Defect Detection System

With the intensification of the production process and the development of competitionin the market, people are increasingly demanding high quality of the products. Therefore,defect detection system based on machine vision to replace manual detection has becomean important measure of quality control for cold-rolled aluminum. This paper mainlydiscusses the research on cold-rolled aluminum surface defect online inspection systemseveral key technologies, including hardware selection，such as linear cameras, lenses,light sources, of the machine vision system for cold-rolled aluminum sheet； andcompositing of the hardware through a variety of experimental methods to the optimalintegration of the program. Software designed for the needs of users, including simple andflexible visual interface, and improved image processing and feature extraction algorithmembedded in software for surface defects. Meanwhile, for the practical applicationenvironment, improve the system software and hardware and validate the system indifferent environments with good reliability and accuracy. Specific studies are as follows:Firstly, the paper illustrates the overall design of the hardware architecture, includingthe selection and installation of the hardware, and hardware integration. In order to meethigh precision detect detection for cold-rolled aluminum, but also taking into account therelatively harsh production environment is not conducive to the operation of electronicproducts for a long time, therefore, the system uses the high-speed linear cameras andhigh-performance IPC were achieved aluminum surface acquisition and processing defects.Gigabit Ethernet was selected as the transport medium after calculation and comparison ofthe actual transmission rate.Secondly, for the high-speed image acquisition is the top-line part to achieve thedefect detection for the aluminum plate. Therefore, to ensure high-speed image acquisition,adjust the portfolio status of each device to continuously optimize image quality, to ensureuniformity of the image, resolution. In the connection with the complex environment ofaluminum production line, ensure the continuity of image acquisition and defect detectionis very importantMeanwhile, the image captured by the high-speed camera array will be fast processed,with distributed software design and IPC array composed with multi-mode of rapidacquisition and processing in the aluminum defect detection system. The processed datawill be sent to sever PC or the on-line PC via Gige-Ethernet road. Further more, the composition also meets the requirements of high-speed image processing, effectively avoiddata congestion.In the software design and implementation process, not only to implement softwaresystems for cameras and other hardware can be a good connection and control, but also beable to perfectly hardware signal receiving and processing, including image signal fromcameras and the encoder pulses. Meanwhile, the system embedded in a variety of improvedimage processing algorithms, first by the image noise reduction and fast median filteringpreprocessing, and then use various thresholds algorithm identified various defects of theimage, such as stripes and the baseline threshold algorithm. Defect is then calculated foreach image region boundary curvature, length and other characteristics of data, the resultsof the calculation area, etc. Furthermore defect classification, statistics defects.Achieved concise interface of the detection system, while ensured the stability of theprogram, and prevented missed and false detection. After detection is complete, theprogram generates the appropriate testing report, the detection results are clear at a glance.The system provides a more intuitive defect data and analysis results for the factory, whichhas brought more good economic benefits.