基于双路图像采集的三维模型构建问题研究

Research on 3D Model Construction Based on Dual Image Acquisition

本文主要研究的是对同一物体的不同方向的两张图片进行分析,并完成曲面重建。从同一的两个不同角度获取两张图像,并对其进行特征匹配处理和密集匹配,最终可以得到一个形成3D点模型,经过适当处理可得模型。关键研究工作内容如下:1)关于SIFT特征匹配的研究应用SIFT特征匹配算法具有更好的优势,因为它能获取更稳定的特征点,并能在图像发生运动时依旧能正确匹配,具有较强的鲁棒性,在目前匹配性能中表现最为突出,在基于特征的匹配中研究最普及的匹配算法。2)M估计抽样一致性算法(MSAC算法)的研究应用利用MSAC算法去除误匹配点。

The purpose of this paper ： Using the camera to obtain two pictures of different directions in the object, and to analyze the image, and finally complete the surface reconstruction. Take two images from two different perspectives of the object, and the feature matching process should performed on two images,after that we should take a dense match on them.As a result,we can get a model that forms by a 3D point cloud.After properly processed,we can get the 3D point cloud model.Research and Application of SIFT Feature Matching Because SIFT feature matching algorithm can get more stable feature points,it has a better advantage than other algorithm,and it work well when object in motion.SIFT algorithm has strong robustness and great matching performance,it is the most popular matching algorithm in feature matching study.Research and Application of MSAC（M-estimator Sample and Consensus） algorithm Using MSAC algorithm to eliminate mistaken matching points.