瞄准摄像机的视觉伺服控制

Visual Servo Control of Gazing Camera

本文提出了瞄准摄像机的视觉伺服控制方法。控制系统由两个模块组成 :图像处理模块和瞄准控制模块。在图像处理模块中 ,用基于特征的方法分析动态序列图像。在瞄准控制模块中 ,用前馈型神经网络实现从信息空间到任务控制空间的映照。根据瞄准摄像机的模型和位姿估计算法通过精确计算得到神经网络的训练样本。用计算机仿真来验证控制方法的性能。

A visual servo control scheme for gazing camera is presented in this paper. The control system is composed of two modules: image processing module and gazing control module. Dynamic sequential images are analyzed by feature-based methods in image processing module. A forward neural network is used to perform the maps from information space to task control space in gazing control module. The training samples of the neural network are obtained from elaborated computation by pose estimation algorithm and gazing camera model. Computer simulations were conducted to verify the performance of the control scheme.