高级检索
陈佳舟, 徐阳辉, 陈坤琦, 罗运学, 缪永伟. 基于镂空标志的保结构增强现实标志隐藏方法[J]. 计算机辅助设计与图形学学报, 2019, 31(11): 1871-1881. DOI: 10.3724/SP.J.1089.2019.17729
引用本文: 陈佳舟, 徐阳辉, 陈坤琦, 罗运学, 缪永伟. 基于镂空标志的保结构增强现实标志隐藏方法[J]. 计算机辅助设计与图形学学报, 2019, 31(11): 1871-1881. DOI: 10.3724/SP.J.1089.2019.17729
Chen Jiazhou, Xu Yanghui, Chen Kunqi, Luo Yunxue, Miao Yongwei. A Structure-Preserving Marker Hiding Approach for Augmented Reality Based on Hollow Markers[J]. Journal of Computer-Aided Design & Computer Graphics, 2019, 31(11): 1871-1881. DOI: 10.3724/SP.J.1089.2019.17729
Citation: Chen Jiazhou, Xu Yanghui, Chen Kunqi, Luo Yunxue, Miao Yongwei. A Structure-Preserving Marker Hiding Approach for Augmented Reality Based on Hollow Markers[J]. Journal of Computer-Aided Design & Computer Graphics, 2019, 31(11): 1871-1881. DOI: 10.3724/SP.J.1089.2019.17729

基于镂空标志的保结构增强现实标志隐藏方法

A Structure-Preserving Marker Hiding Approach for Augmented Reality Based on Hollow Markers

  • 摘要: 标志是室内增强现实应用中广泛使用的相机定标辅助工具,但标志的出现降低了虚实融合的真实感,极大地影响了增强现实画面的和谐度.基于图像修复的标志隐藏方法能够修复去除标志后的画面空洞,但大面积的图像修复往往难以无缝衔接背景的结构特征.为此,设计了一种镂空标志,将传统标志中与定标无关的区域镂空,以减少图像修复区域的面积;并提出一种基于镂空标志的增强现实标志隐藏方法,充分利用镂空标志镂空区域中的背景结构信息,通过一种先修复结构再修复其他区域的保结构图像修复方法,达到快速、无缝的标志隐藏效果.通过ARToolKit增强现实平台和自建测试背景进行实验,结果表明,增强现实标志隐藏方法修复误差小且不影响摄像机的定标精度,弥补了现有方法因大面积图像修复所导致的破坏背景结构特征的不足.

     

    Abstract: The marker is a commonly-used tool for real-time camera calibration in various indoor augmented reality(AR) applications, but the presence of artificial markers reduces the photorealism of the augmented scene, which significantly breaks the harmonious. Marker hiding based on image inpainting techniques repaired the holes where the marker is removed, but inpainting such a large region was too difficult to seamlessly stitch structure features in the background. In this paper, we introduce a novel type of markers called hollow marker, whose calibration-independent regions are hollowed out to reduce the area of image inpainting. And we also propose a marker hiding method based on such hollow markers. This method takes full advantages of the background structures that are revealed in the hollowed region of our markers, produces a fast and seamless marker hiding effect using a structure-preserving image inpainting algorithm. Using the ARToolKit platform and self-built experimental environments, our experiments demonstrate that our approach achieves high quality of marker hiding effects without losing the accuracy of camera calibration, and overcomes the issue of existed AR marker hiding methods that break background structure features due to the large area of image inpainting.

     

/

返回文章
返回