The Research And Implementation Of High Performance Embedded Real-time Image Processing And Recognition System

Along with the science and technology development,the moden industry automaticity level becomes more and more high,and the product line's rhythm becomes more and more quick.Because of the high cost and low efficiency,the traditional quality inspection method already could not satisfy the modern manufacturing industry's demand.The video detection or recognition technology has the merits of non-contact,high accuracy,strong anti-interference,fast speed and high efficiency,and meets the automation producation's requirement.It has become a powerful guarantee of product's quality in modernization production.For example, auto parts automatic assembly,robot vision,industrial detection,and so on.This article details digital image processing algorithm,its performance and results were compared and selected,and its influence to image processing system's real-time. That in turn,use of parallel processing to solve the problem of real-time system is an effective way.Application of FPGA devices for image processing to resolve real-time has the great advantages,by the field programmable gate array(FPGA) structure and working methods of analysis.The article has studied the advanced Altera's DSP Builder development tool in the efficient conduct of Matlab algorithm design and simulation methods,as well as the tool to generate the VHDL code in QuartusⅡand Verilog code of good integration features.The article has analyzed design and workflow of the embedded real-time image processing system,According to the real-time image processing system features, constructed the FPGA-based real-time image processing and identifying of the overall structure of the system,and designed the video capture,video data format conversion, image frame-to-deposit,image processing,real-time display function module.Use of the I~2C bus system for CMOS image sensors and LCD Controller corresponding configuration register.By this way,driven real-time video image capture and display. Video format conversion to achieve the raw data(RAW) format to the RGB format conversion.Image frame buffer is designed for multi-port SDRAM controller,to achieve a single frame or two-frame video images of access to facilitate follow-up method of dealing with the importation of image data.For the largest part of the image processing of the impact of real-time system, Using DSP Builder tool designed for FPGA HW calculation of the structure of the algorithm,has realizated a high-speed image processing median filter,edge detection and image recognition algorithm circuit model.At the same time,the pipeline processing technology and parallel processing techniques applied to the circuit design model,and increased processing speed,the hardware cost savings.Then,in Matlab tool in the design of image processing algorithm simulation debug to observe the simulation environment,the algorithm results.Finally,the algorithm will be designed to deal with the circuit model into the hardware description language to download real-time image processing platform for experiments and analysis of the results.The results show that the embedded image processing and recognition system to meet the system requirements and real-time,has a practical reference value.