多GigE相机高速视觉系统的架构与处理速度研究

Architecture and image processing speed of a vision inspection system with multiple GigE cameras

提出了一种高精度大视场的高速并行处理系统,采用GigE相机并行检测,所有相机排列成线性阵列,通过千兆Ethernet网络与主机连接,并行检测将各单个相机视场合并为一个大视场,在保持单个相机检测分辨率不变的同时扩大检测的范围;并行检测系统可根据图像数据量的变化灵活配置每台相机采集的图像由单台或多台计算机进行处理;数据传输上整个系统采用非碰撞的确定延时网络传输处理结果以确保实时性。通过使用网络延迟理论和OPNET仿真验证得到系统处理的延迟时间不大于162.8μs。

An architecture of vision inspection system is proposed in the paper,which has large field of view（FOV） and can perform high precise,high speed real-time inspection for large objects measurement.The architecture consists of multiple GigE line cameras which connect though Gigabit Ethernet.The cameras are aligned in a horizontal row so that each camera’s FOV is merged into one large FOV in the meantime the resolution keeps unchanged.In order to acquire high processing speed,the captured images from each camera are sent into one computer or multiple com-puters for distributed and parallel image processing.In data transfer process,the technique which adopts non-conflicted data stream with fixed delay is used.At the end system performance is evaluated based on network delay theory and OPNET simulation.It is proved that the whole system process delay is within 162.8 μs.