单幅图像的三维重构视觉系统特征点提取算法的研究与实现  

Research and Realization of the Feature Point Extraction Algorithm in 3D Reconstruction System Using Single Image

作　　者：廖素引[1] 金施群[2] 吴先良[1,3] 马修水[1] 李桂华[1] 

机构地区：[1]安徽大学电子学院,合肥230039 [2]合肥工业大学仪器仪表学院,合肥230009 [3]合肥师范学院,合肥230001

出　　处：《宇航计测技术》2009年第6期12-15,共4页Journal of Astronautic Metrology and Measurement

基　　金：国家自然科学基金资助项目:(No.50275049);(No.06060449)

摘　　要：实现基于单幅二维图像不标定欧氏重构三维场景的关键技术之一是有效提取彩色伪随机编码图像特征角点。为给图像特征点匹配提供更多的样本信息,采用定位于彩色图像Harris算子探测特征角点,并动态给出角点响应的自适应阈值条件。特征点提取实验及三维欧氏重构结果证实了该方法的适用性和有效性。The effective feature point＇s extraction is one of the key technologies in uncalibrated 3D Eaclidean reconstruction based on single pseudorandom color coded image. In order to provide feature matching with more swatch information, introduces Harris＇s color image orientation operator to detect fea- ture point and dynamicly gets an adaptive threshold to judge the corner point. The results of feature extraction experiment and the Euclidean 3D reconstruction confirm the applicability and effectiveness of the present method.

关 键 词：伪随机编码 动态阈值 三维欧氏重构 特征点提取 算法 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]