基于ZYNQ的远程图像采集系统设计

针对传统图像采集系统远程图像传输延时长和数据丢失的缺点，设计了一种基于ZYNQ芯片开发的实时视频采集与图像传输系统；系统具有两个采集通道，模拟视频信号通过BNC(Bayonet Neill-Concelman)信号线接入设备并经过ADC(Analog-to-Digital Converter)完成信号的数字化；利用主控芯片内部的FPGA资源部署并行处理单元完成对数字化图像数据的低时延处理；通过驱动片内AXI(Advanced eXtensible Interface)总线以DMA(Direct Memory Access)的方式将数据传输至DDR3存储器中；利用芯片内部的双核ARM Cortex-A9处理器高性能，在采集设备上移植嵌入式Linux系统，搭建Gstreamer流媒体应用服务器端，实现整个采集系统复杂任务调度和图像数据远程网络传输；与传统单ARM或DSP处理器的图像采集系统相比，该系统具有FPGA(Field Programmable Gate Array)的并行处理能力和高带宽的内部互联总线的优势，提高了图像数据处理速度，降低了图像数据由采集端到存储器的传输延时，提供了稳定远程图像传输功能，经实验测试该系统实现了每秒25帧的视频信号输出，与前端ADC的采集速率保持一致，整个采集展示的过程中视频画面连续且稳定。

Design of Remote Image Acquisition System Based on ZYNQ

Aiming at the shortcomings of long remote image transmission delay and data loss of traditional image acquisition system, a real-time video acquisition and image transmission system based on ZYNQ chip is designed. The system has two acquisition channels, and the analog video signal is connected to the device through the BNC signal line and digitized by the ADC. The FPGA resources inside the main control chip are used to deploy a parallel processing unit to complete the low-latency processing of digital image data; the data is transmitted to the DDR3 memory in the form of DMA by driving the on-chip AXI bus; the dual-core ARM Cortex-A9 processor inside the chip is used to migrate the embedded Linux system to the acquisition device, and the Gstreamer streaming media application server is built to realize the complex task scheduling of the entire acquisition system and the remote network transmission of image data. Compared with the image acquisition system of the traditional single ARM or DSP processor, the system benefits from the parallel processing capability of the FPGA and the high-bandwidth internal interconnection bus, which improves the image data processing speed, reduces the transmission delay of image data from the acquisition end to the memory, and provides a stable remote image transmission function. The experimental test shows that the system realizes the video signal output of 25 frames per second, which is consistent with the acquisition rate of the front-end ADC. The video picture is continuous and stable during the whole process of collection and display.