摘要： 在远程多维空间对象融合过程中,对象之间不一致的空间几何关系会导致观察者空间感错乱,从而造成空间对象融合的沉浸感缺失。为此,基于深度传感器获取的深度信息,克服现有融合方法操作复杂性高、可移植性差及准确率低等问题,提出一种基于几何规则的多维空间对象融合模型。建立背景深度恢复模型修复目标空间的深度缺失区域,设计多层几何表达模型,即几何块结构模型及邻域微结构模型,采用决策树方法恢复源空间前景区域的深度信息,基于彩色信息及深度信息得到几何一致的融合效果。实验结果表明,该模型能够有效降低时间复杂度,同时具有较高的精准度及鲁棒性。

关键词: 空间对象融合, 几何规则, 块结构, 微结构, 深度修复, 几何一致

Abstract: Towards fusing the object from remote multidimensional space,inconsistent geometry relation between objects always leads to confused sense of space,which causes the lack of immersion sense in space object fusion.This paper proposes an object fusion model of multidimensional space based on geometric rule,utilizing the depth information from the depth sensor to overcome the problem of high operation complexity,poor portability and low accuracy in existing fusion methods.It leverages background depth recovery model to repair the area of the target space which is lack of depth information.It establishes multilayer geometric model,that is geometric block-structure model and neighborhood-microstructure model,and uses decision tree method to recover the depth information of foreground in source space.It uses the color and depth information to obtain the geometry consistent fusion effect.Experimental results demonstrate that,the proposed model not only effectively reduces the time complexity,but also has high accuracy and robustness.

Key words: space object fusion, geometric rule, block-structure, microstructure, depth repair, geometric consistency

中图分类号: 

• TP391.3