摘要
针对目前鱼眼相机标定过程繁琐的问题,进行了鱼眼相机快速标定方法的研究。设计了一种由三块黑白棋盘组成、带有红色边界线的立体标定板;对标定图像依据棋盘所属平面进行图像分割,并对每部分分别检测角点,建立二维图像坐标与三维世界坐标的映射;选用折反射成像模型,提出了适用于立体标定法的参数求解方法。进行了该立体标定法与常用平面多图标定法的对比实验,通过图像畸变校正效果和图像坐标逆投影至世界坐标的精度,验证标定精度。实验表明:立体标定法可以实现与平面多图标定方法同样高的标定精度,且具有较好的鲁棒性;同时该标定法极大简化了标定操作流程,缩短了标定时间。
m Aiming at the complexity of current fisheye camera calibration procedures, a study on fast fisheye camera calibration method is carried out in this paper. A stereoscopic calibration board is designed, which comprises three chessboards with red boundaries. After image segmentation according to the planes that different chessboard zones belong to, and running corner detection on each part separately, mapping between 2D image coordinates and 3D world coordinates is achieved. Based on a catadioptric imaging model, a method is introduced to calculate the parameters for stereoscopic calibration. Experiments are conducted to compare this stereoscopic calibration with usual plane-multi-image calibration. The calibration accuracy is evaluated by results of distortion correction and the accuracy of projection from image coordinates to world coordinates. Results indicate that stereoscopic calibration has the same accuracy as that of plane-multi-images calibration and has even greater robustness. Moreover, stereoscopic calibration simplifies the calibration process and shortens the time involved to a great extent.
出处
《哈尔滨工程大学学报》
EI
CAS
CSCD
北大核心
2016年第11期1594-1599,共6页
Journal of Harbin Engineering University
基金
国家高技术研究发展计划项目(2009AA11Z216)
汽车噪声振动和安全技术国家重点实验室开放基金项目(Q10111922)
关键词
鱼眼相机
成像模型
标定方法
标定可靠性
图像分割
参数优化
fisheye camera
imaging model
calibration method
calibration reliability
image segmentation
pa-rameter optimization