基于Legendre神经网络的单目视觉相对位姿解算  

Monocular Vision Relative Position and Attitude Solution Based on Legendre Neural Network

作　　者：郝帅[1] 程咏梅[1] 马旭[1] 赵建涛[1] 崔蓝月[1] 

机构地区：[1]西北工业大学自动化学院,陕西西安710072

出　　处：《微电子学与计算机》2015年第3期41-45,49,共6页Microelectronics & Computer

基　　金：国家自然科学基金项目(60702066);西安市科技计划项目(CXY1350(2));西安科技大学培养基金项目(2014015)

摘　　要：将BP神经网络、RBF神经网络以及Legendre正交基神经网络应用于单目视觉相对位姿强非线性解算,并且在梯度下降法的基础上推导了Legendre正交基神经网络的最优权值,最后进行了相应的仿真及物理实验.实验结果表明,基于Legendre正交基神经网络的单目视觉相对位姿解算方法不仅精度高、鲁棒性强,而且实时性较好.A method of monocular vision relative pose based on Legendre orthogonal basis neural network is proposed.The BP neural network,RBF neural network and Legendre orthogonal basis neural network are combined to solve the strong nonlinear solution in relative position and attitude solution algorithm for monocular vision.The optimal weights in Legendre orthogonal basis neural network are derived from gradient descent method.The developed method can deal with the problem of strong nonlinearity in relative position and attitude solution algorithm for monocular vision.The computer simulations and semi-physical simulations show that the proposed method can obtain satisfactory accuracy,robustness and real-time.

关 键 词：Legendre正交基 神经网络 相对位姿解算 单目 视觉 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]