基于匹配的车载环视图像拼接标定研究

Mosaic Calibration of Vehicle-mounted Surround View Images Based on Matching

车载环视系统是辅助安全驾驶、消除车身四周盲区的重要有效装备,为改善图像拼接标定的精度及实现自动标定,设计图像拼接标定板,提出形状匹配的标定方法。标定板采用圆形图案并按矩形及对角五点布局方式,适用于“F-B-L-R”和“F-B-L-R-Z-Y”及多节牵引挂车场合,基于形状匹配的算法融合最小二乘法、图像金字塔及模板预训练的方法,实现了图像拼接标定。仿真实验表明,标定板具有良好的鲁棒性、算法具有良好的标定精度,能够实现自动化标定。

For the vehicle-mounted surround view system,it is the key and effective equipment to assist safe driving and eliminate the blind area around the vehicle body.The image mosaic calibration board is designed to improve the accuracy of image mosaic calibration and realize automatic calibration,and a calibration method of shape matching is also put forward in this thesis.The calibration plate adopts circular pattern and is arranged according to rectangular and diagonal five points,which is suitable for“F-B-L-R”and“F-B-L-R-Z-Y”and multi-section traction trailer occasions.Then,it realizes the image splicing calibration by combining the least square method,image pyramid and template pre-training based on the shape matching algorithm.According to the simulation experiment,it indicates that the calibration board has good robustness,the algorithm has good calibration accuracy,and is able to achieve the automatic calibration,which lays the foundation for the deep research on vehicle-mounted surround view system.