基于相机标定的全景图拼接方法研究  

Research on Panoramic Image Mosaic Method Based on Camera Calibration

作　　者：马嘉琳[1] 张锦明[1,2] 孙卫新[1] 

机构地区：[1]信息工程大学地理空间信息学院,河南郑州450052 [2]中国科学院遥感与数字地球研究所,北京100081

出　　处：《系统仿真学报》2017年第5期1112-1119,共8页Journal of System Simulation

基　　金：国家自然科学基金(41371383);空间信息智能感知与服务深圳市重点实验室(深圳大学)开放基金资助项目(201501)

摘　　要：提出了一种基于相机标定的全景图拼接方法,对待拼接图像进行几何畸变矫正与旋转变换处理,提高图像配准的精度与全景图的拼接质量。通过相机线性标定得到单应矩阵,分离单应矩阵的内外参数。利用内参数,计算图像的径向畸变,矫正图像的几何畸变;利用外参数,将待拼接的序列图像转化在同一世界坐标系下;利用图像的同名像点实现图像的快速精确配准,实现全景图的高质量拼接。实验结果表明,使用该方法可以获得较高的拼接精度,拼接过程直观且不会发生误匹配。A panoramic mosaic method was proposed based on camera calibration, geometric distortion correction and rotation transformation processing was done for stitching image, in order to improve the accuracy of image registration and panorama stitching quality. The homography was acquired by calculating the camera linear calibration, and the inside and outside parameters of the homography were separated. The radial distortion of the image was calculated by the inside parameters, and the image geometric distortion was corrected,＂ a sequence of images was transformed in the same world coordinates using the external parameters; image registration was achieved accurately though the same name point of a sequence of images, and the high quality of the panoramic image stitching was achieved. Experiment verifies that this method effectively reduces the large amount of calculation in panorama stitching, process of stitching accuracy is higher, intuitive, and matching error will not occur.

关 键 词：全景图 相机标定 图像拼接 图像矫正 图像配准 

分 类 号：P208[天文地球—地图制图学与地理信息工程]