摘要
三维飞行时间摄像机可实时同步获取场景三维信息和灰度图像信息.虽然它存在图像分辨率和质量较差等问题,但它可作为二维摄像机的互补.本文借鉴立体视觉技术,提出了一种2D/3D摄像机融合的三维视觉信息获取方案.论文首先基于固定空间关系和相近视野原则,设计2D/3D立体摄像机系统对空间场景同步成像.结合三维TOF摄像机成像特性,论文借鉴立体视觉技术完成二维摄像机的高质量二维彩色图像与插补后的三维摄像机深度图像的匹配关联.因此,本方法可实现场景的高精度彩色图像和对应三维空间信息的实时同步获取,同时保留了二维摄像机的高质量彩色二维成像和三维摄像机的快速稠密三维信息获取的优势.2D/3D摄像机图像融合匹配算法复杂度低,匹配精度和准确度取决于二维摄像机和三维摄像机自身性能、摄像机标定参数精度和深度图像插补算法,不会引入新的运算误差.试验结果验证了本文算法的有效性和精确度.
Time of flight(ToF) camera can deliver grayscale image and 3D information simultaneously with high frame rates.Although its image resolution and quality are very poor,it provides a good supplement with traditional 2D camera image.Inspired by the stereo vision,we present a 3D visual information acquiring method by fusing 3D information of ToF camera with the high resolution color image.Firstly,we setup a 2D/3D stereo camera with fixed spatial relation and similar visual field to capture the environment information synchronously.Based on characteristics of ToF camera and the principles of stereo vision,the interpolated 3D information of ToF camera is registered with high resolution color image.Consequently,the proposed method can provide high quality 2D image as well as corresponding 3D space information,which preserves the advantages of 2D camera's high quality image and ToF camera's fast acquisition for dense 3D information.The proposed 2D/3D camera fusion method has very low computation cost while the matching accuracy is only determined by parameters of the 2D/3D stereo camera system and interpolation algorithm for range image without extra computation error.Experimental results have demonstrated the feasibility and accuracy of the proposed method.
出处
《控制理论与应用》
EI
CAS
CSCD
北大核心
2014年第10期1383-1392,共10页
Control Theory & Applications
基金
国家自然科学基金资助项目(61005068)
国家高技术研究发展计划("863"计划)资助项目(2012AA112312)
机器人学国家重点实验室开放课题基金资助项目(2013O09)
国家重大科学仪器设备开发专项项目(2013YQ140517)
广东省教育部产学研结合资助项目(2011B090400074)
中央高校基本科研业务费专项资金资助项目