自动多投影仪非线性几何校正与图像边缘融合方法  

Auto-Nonlinear Geometry Calibration and Edge Blending of Multi-Projector Display System

作　　者：王胜正[1] 杨杰[1] 

机构地区：[1]上海交通大学图像处理与模式识别研究所

出　　处：《上海交通大学学报》2008年第4期574-578,583,共6页Journal of Shanghai Jiaotong University

基　　金：国家自然科学基金(60675023);上海市重点学科建设(T0603);上海市教委基金(05FZ12)资助项目

摘　　要：为了解决多投影仪拼接系统中的几何变形与图像融合问题,提出了一种新的自动非线性几何校正和图像边缘融合方法.首先定义了屏幕空间与投影仪之间的几何关系和计算方法;然后用智能相机摄取屏幕上的激光点阵与标准网格投影图像,根据摄取的图像特征计算屏幕空间与投影仪之间的投影变换矩阵,由投影变换矩阵计算得到投影图像的双3次几何变形曲面,从而可以反算得到曲面的控制点,实现对投影图像的非线性几何变形;最后给出图像边缘亮度融合算法.实验结果表明,所提出的方法适用于多投影仪拼接的实时仿真系统.In order to solve the problems of geometric warp and overlapping in a multi-projector display system, a new approach for auto-geometric calibration and edge-blending was presented. First, the homography between screen and projector is defined in this work. Secondly, the intelligent camera is used to obtain the feature points of screen that is defined by laser array and the regular grids that are projected by projectors. The transformation matrices between screen and projector are calculated by the feature extraction of images from the camera. The control points of surface are computed by the bi-cubic surface transformation function that is inferred according to transformation matrices, which implements the non-linear geometric calibration of projected images. Finally, the edge-blending function is presented. The experimental results illustrate the feasibility of the proposed method.

关 键 词：自动几何校正 多投影仪 图像边缘融合 虚拟现实 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]