| DPM(Direct Part Marking) is 1D or 2D marking which is permanently marked on object’s surface. DPM can describe and store many kinds of information about the object,which is very suitable to achieve the goal of Identi?cation and Traceability proposed by ISO(International Organization for Standardization) 9000 standards.Automated Optical Recognition is a technology that can automatically obtain optical images and recognize information by means of Machine Vision. Automatic Identi?cation and Data Capture is a technology that can capture and identify information of objects without any manual intervention. These two technologies should be used together with other technologies in order to design DPM automated optical recognition system.DPM was ?rst used in 1980 s by foreign manufacturing industries such as automotive, aerospace, electronics and semiconductor. At present, domestic research of DPM are mainly focused on how to mark DPM onto different substrates, the research of DPM automated optical recognition system is rarely reported. Most DPM recognition systems use hardware and software developed by foreign researchers and companies. Relevant research area is Optical Mark Recognition(OMR). However, OMR systems are not optimized for DPM, and most of them require a PC(Personal Computer) to implement recognition algorithms. As DPM is more and more widely used, the research and development of DPM automated optical recognition system is imminent.The research background is based on DPM automated optical recognition system design. This thesis focused on 2D-form DPM which is widely used in our country, and?nally an embedded system architecture based on FPGA and DSP devices was designed.After basic principles and overall design plan of this system were described, the hardware and software were accomplished by means of different strategies.The hardware was designed using bottom-up strategy. Starting from chip level, the whole hardware platform was built up. CMOS image sensor is used to capture optical images of DPM. FPGA is a kind of highly programmable devices, it is convenient to recon?gure it to get updated and upgraded according to actual needs, thus increasing the?exibility and scalability of this system. DSP is specially optimized for digital signal processing, in this system ADSP BF60 x was used. As far as we know, this thesis is the?rst report of using it in cooperation with FPGA to automated optical recognize DPM in China. Other hardware modules include external memory, Ethernet communication and power-supply.The software was designed using top-down strategy. Starting from system function,the whole software platform was split into four modules. They are TWI serial communication module, optical image capture and transfer module, DPM recognition module and Ethernet communication module. lw IP(a open-source light-weight TCP/IP stack) was transplanted to achieve fast Ethernet realtime communication.After hardware and software platform were built up, joint debugging was conducted.Experimental results show that this embedded system has achieved the intended functions, the speed of 2D-form DPM automated optical recognition is about 60 per second and accuracy rate is above 99%. |
