视差图提取技术与OpenGL三维重建  

Parallax figure extraction and OpenGL 3D reconstruction

作　　者：罗桂娥[1] 陈少华[1] 刘小群[1] 

机构地区：[1]中南大学信息科学与工程学院,湖南长沙410083

出　　处：《微型机与应用》2012年第9期38-40,共3页Microcomputer & Its Applications

摘　　要：在研究区域匹配算法和特征匹配算法的基础上,提出了改进的基于视差梯度的区域匹配算法和基于尺度不变性的Harris角点特征匹配算法,并进一步利用互补策略将两种算法结合起来,提出了一种区域和特征匹配相结合的立体匹配算法,该算法具有速度快、精度高和鲁棒性强等优点。利用该算法提取视差图,进而提取深度图,最后利用OpenGL进行三维重建,获得了良好的重建效果。Based on the research of area-based stereo matching algorithm and feature-based stereo matching algorithm, this paper introduces area-based stereo matching algorithm based on disparity gradient and feature-based stereo matching algorithm based on Harris comer of scale invariance. In further, this paper combines the two algorithm with the use of complementary strategy to propose a stereo matching algorithm, which has the characteristics of high speed, high precision and strong robustness. Using this algorithm, the paper extracts parallax figure and then extracts depth map. At last, three-dimension image using OpenGL on computer is rebuilt, and it obtains good effect.

关 键 词：视差梯度 尺度不变性 互补策略 OPENGL 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]