基于Kinect的深度数据融合方法

与传统三维激光扫描仪相比, Kinect作为一种新型深度相机, 具有价格低廉、深度数据获取能力强、RGB影像与深度影像同步获取等优势, 然而面对较大室内场景精细建模却存在数据量大、建模范围有限、对硬件环境依赖性强等问题。因此, 在现有单一模型建模基础上, 提出了基于Kinect深度影像的多模型数据融合方法, 实现模型间的自动拼接。最后通过两组实验对提出的数据融合方法进行了验证, 并取得了较好的模型拼接效果。

Kinect-based depth image fusion method

Compared with traditional 3D scanners, Kinect is a new kind of range camera, it has advantages of low price, strong capability of data collection, synchronous acquisition of RGB image and depth image. However, it exists problems of large amount of data, limited reconstruction range, and high dependence on hardware, when Kinect is used to build large and fine indoor scenario model. Therefore, this paper proposed an effective solution to reconstruct high-resolution 3D model in large indoor space and it was especially to solve the problem of data fusion with different Kinect image data in the same scene and merging with different models automatically. Finally, two sets of experiments verified the proposed method and the result proved that it was very effective.