基于单应性矩阵追踪的自动焦距标定算法  

Auto focal length calibration algorithm based on homography matrix tracking

作　　者：孙翌乔 王春勇[1] 李振华[1] 张宝辉 Sun Yiqiao;Wang Chunyong;Li Zhenhua;Zhang Baohui(School of Science,Nanjing University of Science and Technology,Nanjing 210094,China;Kunming Institute of Physics,Kunming 650223,China)

机构地区：[1]南京理工大学理学院,江苏南京210094 [2]昆明物理研究所,云南昆明650223

出　　处：《南京理工大学学报》2022年第2期233-238,共6页Journal of Nanjing University of Science and Technology

基　　金：国家自然科学基金(61971225);数字成像国防基础科研项目(JCKY2016606B001)。

摘　　要：为了减少旋转式全景拼接系统中由于焦距不准确导致的图像畸变问题,该文在传统焦距标定算法的基础上,提出了一种有针对性的焦距标定算法。该算法对依照不同焦距进行柱面投影时相邻图像间的单应性矩阵参数进行追踪,通过二次函数拟合确定最佳焦距,实现了焦距的精标定,保证了360°全景拼接时不出现较大畸变。经该算法处理后的全景图像畸变明显小于现有算法。In order to reduce the image distortion caused by inaccurate focal length in the rotary panoramic mosaic system,this paper proposes a targeted focal length calibration algorithm based on the traditional focal length calibration algorithm.In this algorithm,the homography matrix parameters between adjacent images are tracked when cylindrical projection is carried out according to different focal lengths,and the optimal focal length is determined by quadratic function fitting.The precise calibration of focal length is realized,which ensures that there is no large distortion in 360°panoramic mosaic.The distortion of panoramic image processed by this algorithm is obviously less than that of the existing algorithms.

关 键 词：全景成像 图像拼接 焦距 柱面投影 图像处理 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]