Research Of Image Acquisition And Pre-processing System Of LED Packaging Process Intelligentization

With the development of LED packaging technology,the LED downstream industry of china has been forming larger market sizing,but the bottleneck of LED product yield is still hard to break through.LED packaging process intelligentize increases the defect detection and process backtracking mechanism on the basis of existing technological process,when quality fluctuates,positioning and eliminating problems through backtracking production process information of defect LED,which is aimed at improve LED product yield.Defect detection and process backtracking mechanism use of image processing techniques,faced with mass LED produced with high speed,the realtime performance bottleneck of image processing is one of the main barriers in realizing intelligent manufacturing.Image acquisition and image preprocessing are premise of image processing and image analyzing,the paper research on the hardware acceleration of image acquisition and image preprocessing based on FPGA parallel processor,to improve the realtime performance of defect detection and process backtracking mechanism.Main research contents of this article are as follows:(1)Have proposed a general method of obtaining operation display images of equipment by collecting VGA display signal.This method is able to overcome difficuty of production data acquisition directly because LED packaging equipments lack of data interface;(2)Have studied the technical scheme of image acquisition system,have designed VGA display image acquisition board and CMOS image sensor acquisition board with FPGA core controller,have analyzed the shortcomings and improvement schemes;(3)Have studied the scheme of controlling peripheral of FPGA,have implemented I2 C controller module,image acquisition module,SDRAM controller module,UART controller module and image display module by Verilog.Besides,have verified functional correctness of all modules by Modelsim simulation tool;(4)Have studied the hardware implementation principle of image preprocessing algorithms,have designed and verified line cache module,Gauss filtering module,median filtering module and Roberts,Laplace,Sobel edge detection algorithms modules;(5)Have proposed a fast median filtering circuit with complete parallel structure and a improved sobel edge detection circuit with gradient non-maxium suppression function.Besides,have optimized the architecture of Gauss filtering and Sobel edge detection modules,which made less hardware resource consumption;(6)Have tested the effect of all image preprocessing algorithm modules with hardware realization and estimated the time-consuming of image preprocessing algorithms which were realized in different processors,the contrast result of uptime reflected the real-time advantage of FPGA in image processing application.