基于多观测点图像SURF特征配准及去模糊的三维重建  

Multiple Observed Images Deblurring and 3D Reconstruction Based on SURF Feature

作　　者：时愈[1] 罗枭[1] 邓哲煜 曹太腾 范艳[1] 

机构地区：[1]武汉工程大学电气信息学院图像处理与智能控制研究室,湖北武汉430205

出　　处：《光学与光电技术》2016年第1期28-32,共5页Optics & Optoelectronic Technology

基　　金：国家自然科学基金面上项目(61175013);湖北省自然科学基金创新项目(2012FFA046);武汉市国际科技合作计划项目(2014030709020310)资助项目

摘　　要：动目标多观测点图像去模糊及三维重建是三维视觉检测与测量技术应用中的难题,而特征检测对去模糊及三维重建的结果影响较大。针对这个问题,提出了一种基于多观测点图像SURF特征配准及去模糊的三维重建方法。首先对图像进行SURF特征点检测并对这些特征点进行配准,根据配准的特征点求解Kruppa方程得到各视点图像的相机内外参数矩阵,进而求取图像的点扩散函数即模糊核并对图像进行去模糊处理。其次,提取图像中的SURF特征点并进行配准,求取任意两幅图像的仿射变换矩阵,获取多观测图像的像素点投影。最后根据SURF特征的配准及多观测的投影结果,对去模糊后的图像进行立体匹配,从而完成多观测图像的三维重建。实验结果表明提出的方法对多观测点图像去模糊及三维重建具有良好的效果。Multiple observed image deblurring and 3Dreconstruction is a difficult problem in the application of three-dimensional vision inspection,and feature detection is important for deblurring and 3Dreconstruction.In order to address this problem,a multiple observed images deblurring and 3Dreconstruction method based on SURF feature was proposed in this paper.Firstly,the parameter matrixes of cameras are obtained by Kruppa equation by SURF feature detection.Then the point spread functions are obtained and images are deblurred.Secondly,the affine transform matrixes and projection results are obtained by feature detection.Finally,according to matching and projection results of SURF feature,the stereo matching for 3Dreconstruction with deblurred multiple observed images is implemented.A series of experiments show the effectiveness of the proposed method.

关 键 词：多观测 运动模糊 SURF特征 特征配准 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]