Implementation And Algorithm Applications Of Image Processing System Based On PL-PS Architecture

Video processing has more and more important status in communication, production management, multimedia technology, high-definition television, medical, target tracking, and other fields.The traditional video processing platform widely used PC, application-specific integrated circuit chip or DSP device, and require a high frame rate as far as possible.Due to the parallel structure of the FPGA, it has the advantages of processing speed,and the power consumption is low, so it is more and more applied to the video processing field in recent years. However, simply using FPGA for video processing has its limits, the implementation of high-level algorithms by FPGA, such as machine learning, scenario analysis, need to be combined with FPGA hardware circuit characteristics,then we should reorganize the algorithm, so that the realization of the algorithm will be more difficult, the consumption of the on-chip resources will be much more, and the R&D cycle will be long.In industrial and academic fields, a common video processing system scheme is using FPGA to capture and preprocess, and sends the data to PC through the PCIE interface, then PC further process and analyze the data. However, this scheme also had a disadvantage of large volume, high cost, for the hot field of automotive electronics, security, the scheme is not a good solution.This paper achieve a use video processing system by the development board based on Zynq chip. Zynq chip,with PL-PS chip architecture,was launched by Xilinx company in 2011. The PL means Programmable Logic,that is, the FPGA; the PS means Processing System, that is, the ARM. The tiny chip, with the integration of 28nm FPGA and a dual-core Cortex A9 ARM,could use AXI bus to communicate between PL and PS, and both PL-PS can share one piece of memory, co-design by software and hardware. The chip could take advantages of both FPGA and ARM,and the FPFA and ARM can be complementary each other.For some application (such as video processing),this architecture can bring performance improvement.In this paper, the video processing system based on the Zynq chip can realize the function of video capturing, processing, storage and displaying. With the parallel characteristic of FPGA device, the system can realize image preprocessing in PL,for instance,color format conversion, denoising operations. In PS,because of its serial processing characteristicsand can be embedded operating system, It can realize more complex operations, can achieve advanced process of image, such as characteristic analysis, classifier and so on. in addition, PS can also do some control operation, such as the servo control, etc. Through the way of the software and hardware co-design,it can improve the performance of video applications.In addition to the achievement of data flow path, this system also achieve several image algorithms based on FPGA, such as camera raw image saw tooth removal algorithm, Sobel filter algorithm, RGB and HSV color space conversion algorithm. The PS part of the system can be embedded with Linux operating system or run at standalone mode(without operation system).There are some innovations in this system.1. The innovation is in architecture. The system adopts the Zynq chip, which integrates with FPGA and ARM processor, and each has its advantage. The FPGA-ARM architecture promote the performance.2. The modular design method is adopted in the hardware design of this system, each function is packed into IP, and at last, they connect each other by dragging the wire by mouse. The method is convenient and improve the efficiency of the development.3. The application of HLS tool, make the software engineers could develop the FPGA applications without the knowledge of verilog HDL, so this can decrease the difficulty of FPGA development, and this can improve the development cycle. Because of the Zynq chip performance advantages, in the future, the Zynq chip will play a more important role in the field of video processing.In this paper, the video processing system, the project content includes the FPGA and ARM device, also includes the image processing algorithms, which relate to a wide range of knowledge, so for the same industry developers this paper has a certain reference significance.