基于相机参数标定的全景监控影像拼接方法  

A Panoramic Surveillance Image Stitching Method Based on Camera Calibration Parameters

作　　者：朱格 李冰冰 曹晗 李峰[1] 潘雨青[1] ZHU Ge;LI Bingbing;CAO Han;LI Feng;PAN Yuqing(Jiangsu University,College of Computer Science and Information Engineering,Zhenjiang Jiangsu 212013)

机构地区：[1]江苏大学计算机科学与信息工程学院,江苏镇江212013

出　　处：《软件》2023年第9期78-83,179,共7页Software

摘　　要：为了提高全景监控视频的实时性,本文提出了一种基于相机标定参数的图像拼接方法。为了获取360°全景图像,该方法采用了6个固定角度的摄像头,并通过棋盘格标定板对摄像头的内外参数进行精准标定,构建出6个相机坐标系与世界坐标系的关系模型。然后通过球面投影模型,将画面投影到拼接平面上,最终达到全景监控画面的拼接效果。另外,为了更好地减少相机受外力干扰的影响,我们引入了一种参数维护机制。通过低功耗嵌入式平台和拼接服务器相结合的实验,表明该方法能够有效地提高图像拼接的速度和精度,不受环境光照条件和特征点匹配的影响,适用于无特征点或环境变化大的场景下的实时图像拼接。To improve the real-time performance of panoramic surveillance video,this paper proposes an image stitching method based on camera calibration parameters.To obtain a 360°panoramic image,this method uses six cameras with fixed angles,and calibrates the internal and external parameters of the cameras accurately by using a chessboard calibration board,and builds a relationship model between six camera coordinate systems and the world coordinate system.Then,through the relationship of spherical projection,the images are projected onto the stitching plane,and finally achieve the stitching effect of panoramic surveillance images.In addition,in order to better reduce the impact of external forces on the camera,this paper introduces a parameter maintenance mechanism.The experimental verification based on the combination of low-power embedded platform and stitching server shows that this method can effectively improve the speed and accuracy of image stitching,and is not affected by environmental lighting conditions and feature point matching,and is suitable for real-time image stitching in scenes with no feature points or large environmental changes.

关 键 词：图像拼接 全景视频 视频拼接 监控摄像头 相机标定 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]