Location:
Search - Calibrat
Search list
Description: 介绍了基于模型的位姿估计中所使用的一些优化方法。为了提高位姿估计的精度, 摄像机的标定参数必须足够精确,
这就对标定过程的非线性优化算法提出了很高的要求, 采用了一种新的优化目标函数, 用来最小化控制点间的三维重建误
差, 从而使标定参数是全局最优 在双像机位姿估计中, 引入了实时遗传算法进行全局搜索, 加快了算法的收敛速度。最后的
实验证明了这些方法的正确性并显示出这些方法在精度上比传统方法有了较大程度的提高- It int roduces s ome opt imizat ion methods in model-based pose est imat ion. Camera calibration parameters
mus t be accurat e enough t o guarantee est imation precison. A new cos t function is proposed t o minimize the
3D reconst ruct ion error of control point s, and it makes calibrat ion paramet ers globally optimal. In binocular
case, real-time genetic algorithm( RGA) is ex plored f or global searching, and it improves the converg ent speed.
Ex periment s verif y these schemes , and they can improve es timat ion precis ion dramatically.
Platform: |
Size: 111616 |
Author: cp |
Hits:
Description: 提出了微惯性测量单元( MIMU) 在高动态、高过载复杂应用条件下的误差整机标定和补偿方法。首先, 建立了高动态, 高过载复杂应用条件下MIMU 的误差模型, 该模型包括了结构误差, 传感器安装误差和MEMS 惯性传感器在复杂条件对精度影响较大的误差项, 指零位温度漂移、互耦误差、刻度因子非线性和微陀螺加速度效应误差; 根据模型提出了整机标定补偿方法, 该方法可以标定MIMU 的63 个误差系数, 并且不需要对单个传感器进行标定。然后, 介绍了利用最小二乘法对模型进行误差系数标定的方法和步骤, 并对自研的MIMU 进行了标定。(T he entire calibrat ion and compensat ion metho d o f a Miniature Inerial Measur ement U nit( MIMU) in high dy namic and o verload complicate environments w as pr opo sed. First ly, an error model applied to the complicate applicat ion environment s w as established, which consist s o f the structureer rors, installatio n misalignment errors and the erro rs of the MEMS sensors including zero outputdrif t, temperature drif t , crossax is error, no nlinear scale factor erro r and acceler at ion effect err or of
gyr oscope. Based o n the model, the ent ire cal ibrat io n and compensat ion method w as pr opo sed to calibrate 63 er ror co ef ficient s w ithout calibrat ion of each MEMS iner tial sensor separ ately. T hen, thegeneralized least square algor ithm was used to calibrate and calculate the er ror coef ficient s. Finally, a
MIMU was dev elo ped for a f light ex periment and w as calibrated w ith this pro posed method.)
Platform: |
Size: 539648 |
Author: guyan
|
Hits: