摘要
针对多相机协同的隧道快速检测存在分布在多张图像中的一个病害误识别为多个病害影响隧道技术状况评定的问题,提出一种数据和场景驱动的多相机序列图像高精度拼接方法。首先,利用场景中相机间的几何位置关系进行几何粗解算,生成序列图像的理论拼接模式,其次,基于理论拼接结果计算图像关系,对相邻重叠图像采用SURF算法提取特征点并进行匹配,进行像素级的数据配准,最后,基于配准结果生成实际拼接模式,以最大化利用图像物理分辨率高的数据为特征,在拼接序列图像中提取断面数据。实际工程数据表明,该方法能够实现普通隧道的图像拼接工作,并且能在保证可靠性的前提下尽量提高准确性。
In the rapid tunnel detection of multi-camera,a disease distributed in multiple images is misidentified as multiple diseases affecting the evaluation of the technical status of the tunnel.This paper proposes a data-and scene-driven multi-camera sequence image high-precision stitching method.First,geometric rough calculation is done to generate the theoretical stitching mode of sequence images using the geometric positional relationship between the cameras in the scene.Secondly,the image relationship is calculated based on the theoretical stitching results.Feature points are extracted for adjacent overlapping images through SURF algorithm and matched.Pixel-level data registration is performed.Finally,an actual stitching mode is proposed based on the registration results.The feature is to maximize the use of data with high physical resolution of the image to extract cross-section data in the stitched sequence image.Practical results show that this method can achieve the image stitching work of ordinary tunnels,and can maximize the accuracy under the premise of ensuring reliability.
作者
王照远
曹民
王毅
吴伟迪
李陶胜
WANG Zhaoyuan;CAO Min;WANG Yi;WU Weidi;LI Taosheng(School of Electrical and Electronic Engineering,Hubei Univ.of Tech.,Wuhan 430068,China;ZOYON Company Limited,Wuhan 430223,China;Anqing Vocational and Technical College,Anqing 246003,China)
出处
《湖北工业大学学报》
2020年第4期11-15,共5页
Journal of Hubei University of Technology
基金
武汉市科技计划项目(2016070204020160)
安徽省教育厅高校自然科学研究重点项目(KJ2019A1193)。
关键词
公路运输
隧道检测
图像拼接
场景配准
SURF算法
road transport
tunnel detection
image stitching
scene registration
SURF algorithm
