【作者】 曾慧； 穆志纯； 袁立；

【Author】 ZENG Hui MU Zhi-chun YUAN Li(School of Information Engineering & University of Science and Technology Beijing,Beijing 100083)

【机构】 北京科技大学信息工程学院；

【摘要】 目前,三维人耳信息的获取主要是通过激光扫描仪来获取的,在基于二维图像的三维人耳重建技术方面,还没有成熟的理论和方法可供使用。从我们检索的文献来看,仅仅有 David Zhang 的研究小组提出的一种人工交互式的基于图像的三维人耳重建方法。针对这一情况,本文提出了一种全自动地三维人耳重建方法,主要包括人耳图像 SIFT 特征点提取与匹配、特征点扩散、摄像机自标定和三维重建等步骤。实验结果表明,该方法能够全自动地、有效地重建出较为稠密的三维人耳空间点。更多还原

【Abstract】 Now 3D ear information is obtained mainly by laser scanner and no state-of-the-art 2D image based 3D ear reconstruction method is presented.As far as we know,the only one 2D image based 3D ear reconstruction method is proposed by the research group of David Zhang and it needs human interaction.In this paper,we present an automatic 3D ear reconstruction method.At first,the seed matches are obtained by SIFT.Then the dense point correspondences are gotten by match propagation.Finally we perform camera self-calibration and computed the 3D ear points.Detail Experimental results are shown that this method can obtain dense 3D ear points automatically and effectively.更多还原