基于RGB⁃D图像的室内场景三维布局规划模型构建  

Construction of indoor scene 3D layout planning model based on RGB⁃D image

作　　者：周江[1] ZHOU Jiang(Shanghai Normal University Tianhua College,Shanghai 201815,China)

机构地区：[1]上海师范大学天华学院,上海201815

出　　处：《现代电子技术》2021年第4期138-141,共4页Modern Electronics Technique

基　　金：国家自然科学基金项目(61203108)。

摘　　要：传统的基于Scene⁃SIRFS算法的室内场景三维布局规划模型受到室内遮挡物的影响,造成物体几何信息提取不准确,为此设计一种基于RGB⁃D图像的室内场景三维布局规划模型。依据室内场景图像平面检测,得到同一个物理面中的多个室内空间子面,通过多个室内空间子面中的特征点匹配,确定室内场景中物体布局位置,为准确对场景中物体几何特征描述,添加了遮挡类边缘约束条件,并依据RGB⁃D图像的全局能力,对室内场景三维布局规划目标函数求解,完成室内场景三维布局规划模型的构建。实验以提取桌子的几何特征信息为实验对象,结果表明,由此次设计的模型得到的物体几何边缘信息比传统模型得到的物体几何边缘信息更加完整、清晰,证明此次设计的模型比传统模型提取三维几何信息时的准确度高。As the traditional indoor scene 3D layout planning model based on Scene⁃SIRFS algorithm is affected by the indoor obstructions,which causes the inaccurate extraction of geometric information of objects,a indoor scene 3D layout planning model based on RGB⁃D image is designed.The multiple indoor space sub⁃surfaces in the same physical plane are obtained according to the indoor scene image plane detection.By matching the feature points in multiple indoor space sub⁃surfaces,the layout position of objects in the indoor scene is determined.In order to accurately describe the geometric characteristics of objects in the scene,an occlusion⁃type edge constraint is added,and the 3D layout planning objective function of the indoor scene is solved based on the global capabilities of the RGB⁃D image to complete the construction of the 3D layout planning model of the indoor scene.The experiment takes the geometric feature information of the table as the experimental object.The results show that the object geometric edge information obtained by the designed model is more complete and clear than the object geometric edge information obtained by the traditional model,which verifies that the designed model is more accurate than the traditional model when extracting the 3D geometric information.

关 键 词：室内场景 三维布局规划模型 RGB⁃D图像 场景解析 特征点匹配 模型构建 

分 类 号：TN911.73-34[电子电信—通信与信息系统] TP311[电子电信—信息与通信工程]