Research And Application Of The Machine Vision Based On FPGA

Machine vision is a rapid, objective, economic and non-contact inspection tool. In recentyears, machine vision technology has been widely used in industrial inspection. However,with the development of the industry automation, the demands of industry inspection becomemore and more strict. Traditional machine vision products based on PC can hardly fulfill thedemand of high speed and real-time system.With the development of microelectronics technology, FPGA hardware resources have agradually increasing. FPGA becomes more competitive because of the advantage of hardwareresources configuration, parallel computing, high-speed, low-power consumption, low-costand so on. This thesis combined FPGA technology with visual processing technology that cantake advantage of hardware implementation, parallel processing, multi-stage pipelinetechnology which can improve data throughput capacity and data processing capacity inmachine vision inspection processing system. FPGA-based machine vision technology is animportant research direction in machine vision technology.With the Yarn textile speed up, the traditional detection of broken yarn can hardly fulfillthe real-time requirements. Aiming at the real-time detection of broken yarn, this thesisdeveloped an FPGA-based the broken yarn visual detection system.Firstly, this thesis designed the broken yarn visual detection platform. The platformincludes a linear array CCD camera, a LED strip light source, a FPGA board. Modules weredesigned responsible for data collection, storage, processing and VGA display.Secondly, two different kinds of detection algorithms were proposed for the broken yarnvisual detection and were implemented by using the VHDL hardware description language inFPGA. The algorithms which use the hardware resources configuration, parallel processing,pipeline processing have a good performance in parallel data processing and real-time system.The algorithms were tested in a real production environment; test results shows that thedetection algorithm based on the projection fulfilled the demand of real-time broken yarndetection.Finally, aiming at the disadvantage of system in FPGA-PC data communication ability,this thesis designed a network communication module for the broken yarn system. In order toreduce consumption of the hardware resources, thesis had a wide range cut of the TCP/IPprotocol stack. A FPGA used as the main controller and a Ethernet controller chip DM9000Aused as the MAC layer was implemented the simple TCP/IP protocol stack in which is thenetwork communication module The module design used the SignalTap II debugging tool andWireShark network listener tool to have the function test, test result demonstrated that thenetwork communication module works stable, dada transmit speed reached to7.5MB/s.