Research And Implementation Of Character Recognition System Based On Zynq-7000

With the continuous deepening of the machine vision technology, applications which is related with embedded machine vision have penetrated into all aspects of our daily life. Such as automatic driving car, intelligent game, as well as some industrial applications of automation equipment, are embedded in specific applications of machine vision. However,due to the limitations of traditional PC computing power, a lot of real-life scenarios of application(real-time HD image processing, high frame rate real-time video processing) are very difficult to meet the requirements of real-time processing, and traditional design methods has failed to meet the product for miniaturization, low power consumption and processing large image data sets of requirements.The design is based on Xilinx Zynq-7000 platform, through the method of hardware and software co-design, completed the design of embedded character recognition systems to meet the definition of a character image processing real-time requirements. The design work is mainly done several of the following areas:Firstly, it introduces the basic module of character recognition system, and the processing flow of the whole character recognition system, and completes the algorithm implementation of each module.Secondly, the hardware architecture of Zynq platform is introduced, and the specific platform module of character recognition system is described in detail. And the embedded Linux operating system is built, including the transplantation of U-boot, the kernel of the compiler and the construction of the file system. The transplantation of open source computer vision library is realized.Thirdly, this paper focuses on the Xilinx hardware and software collaborative development tools Vivado HLS, including input and output file of HLS composition, HLS function library and HLS project development process, and introduces the Zynq SoC video processing architecture and the specific video processing process. And OpenCV and HLS were used to realize the pretreatment module, and the pretreatment of HLS was analyzed.Finally,the software and hardware development process of the whole character recognition system is described in detail, and the whole system is tested and the results are given.