基于单一图像实现城市场景下的建筑三维布局
    点此下载全文
引用本文:刘卓亚?覮.基于单一图像实现城市场景下的建筑三维布局[J].计算技术与自动化,2019,(1):160-166
摘要点击次数: 633
全文下载次数: 0
作者单位
刘卓亚?覮 (云南机电职业技术学院 工业信息技术系云南 昆明 650203) 
中文摘要:旅游资源和信息技术的有机结合能够释放巨大的商业潜力,具有重要的社会意义。从一个单一的二维图像的旅游城市场景推断建筑立面的三维布局。与现有的只产生粗方向标签或定性块近似的方法不同,所提出的算法使用三维几何约束相互关联的一组平面定量地重建三维空间中的建筑立面。每个平面的特征是连续的方向矢量和深度分布。通过平面间的相互作用来达到最佳的解决方案,由于该算法几何推理的定量和基于平面的性质,所提出的算法模型比现有的方法更具表现力和更丰富的信息。实验结果表明,本文的方法与采用测量方法实现三维重构方法相比,能够更加逼真的重现三维场景。
中文关键词:三维重构  建筑立面  几何建模  约束条件  图像深度估算
 
Realizing 3D Layout of Buildings in Urban Scene Based on Single Image
Abstract:The organic combination of tourism resources and information technology can release huge commercial potential and has important social significance. This paper infers the three-dimensional layout of the building facade from a single 2D image of the tourist city scene. Unlike existing methods that only produce coarse-directed labels or qualitative block approximations,our algorithm uses a set of planes that are related to each other by three-dimensional geometric constraints to quantitatively reconstruct building facades in three-dimensional space. The features of each plane are continuous directional vectors and depth distributions. Through the interaction between planes to achieve the best solution,due to the quantitative and plane-based nature of our geometric reasoning,our model is more expressive and informative than the existing methods. The experimental results show that the method in this paper is more realistic than the three-dimensional reconstruction method using measurement methods.
keywords:3D reconstruction  building facades  geometric modeling  constraints  image depth estimation
查看全文   查看/发表评论   下载pdf阅读器