摘要
为了从物体的二维图像得到三维重构模型,需要通过相机内部参数,建立已知物点、像点对应的关系模型.提出一种基于双平面镜的相机标定算法,用两个普通平面镜取得物体5个不同角度的二维图像,通过基于颜色信息和基于区域背景差的方法获取目标轮廓,用多边形动态规划算法获取轮廓的特征点以确定各个物体轮廓的对应点,最后根据特征点得到相机内部参数.多边形动态规划算法将时间复杂度从O(n3)降低到O(n2),算法的效率得到提高.
Object＇ s 3D reconstruction parameters are retrieved by the relationship between object＇s feature points in the scene with their correspondence in 2D images taken by calibrated camera. A camera calibration method based on two-mirror set is presented for parameter computation. With two-mirror set, an object is photographed from 5 different directions in one image. Visual contour is calculated by background subtraction based on color and regional information. A polygon dynamic programming is used to determine feature points on the visual contour and then their correspondences. Camera parameters are calculated with the constraints of the feature points with their correspondences. With polygon dynamic programming, and the time complexity is reduced from O（n2） to O（n2） , the efficiency is improved.
出处
《东华大学学报（自然科学版）》
CAS
CSCD
北大核心
2012年第1期93-98,共6页
Journal of Donghua University(Natural Science)
关键词
背景差
轮廓提取
多边形凸点
三维重构
相机标定
background subtraction
contour extraction
convex hull
3D reconstruction
camera calibration
