發(fā)布時間：2018-12-13 06:30

【摘要】：針對車載全景序列影像的位姿估計問題,該文提出了一種基于角元素和線元素的位姿估計優(yōu)化方法。根據(jù)全景多視幾何關(guān)系,建立了角元素和線元素多視約束的優(yōu)化模型,并推導(dǎo)完成模型的位姿求解方法。然后作為初始外方位元素,加入稀疏GPS/INS的相機控制數(shù)據(jù)約束,采用全景球模型的稀疏光束法平差對位姿估計結(jié)果進一步優(yōu)化。實驗選擇IP-S2車載移動測量系統(tǒng)獲取的全景序列影像,對比分析單視圖、全景視圖的連續(xù)位姿估計、本文方法和GPS/INS控制數(shù)據(jù)。實驗結(jié)果表明,該文方法可提高位姿估計的精度,并減少連續(xù)位姿估計產(chǎn)生的累計誤差。
[Abstract]:Aiming at the problem of pose estimation of vehicle panoramic sequence images, this paper proposes an optimal method for position and pose estimation based on angle element and line element. According to the panoramic multi-view geometry, the optimization model of angle element and line element multi-view constraint is established, and the method to solve the position and pose of the model is deduced. Then as the initial external azimuth element, the camera control data constraints of sparse GPS/INS are added to optimize the alignment estimation results using the sparse beam adjustment of the panoramic sphere model. The panoramic sequence images obtained by IP-S2 vehicle mobile measurement system are selected in the experiment, and the continuous pose estimation of single view and panoramic view is compared and analyzed. The method of this paper and the control data of GPS/INS are compared and analyzed. The experimental results show that the proposed method can improve the accuracy of pose estimation and reduce the cumulative error caused by continuous pose estimation.
【作者單位】： 遼寧工程技術(shù)大學(xué);
【基金】：國家自然科學(xué)基金項目(41501504)
【分類號】：TP391.41;U463.6

【相似文獻】

相關(guān)期刊論文 前1條

1 羅勇;陳慧;;基于EKF的自動泊車系統(tǒng)位姿估計算法設(shè)計[J];山東交通學(xué)院學(xué)報;2013年04期

，

本文編號：2376074