计算机与现代化

• 图像处理 • 上一篇    下一篇

 基于RGB-D序列的人体动态建模方法

  

  1. 北京工业大学,北京100124
  • 收稿日期:2014-03-24 出版日期:2014-06-13 发布日期:2014-06-25
  • 作者简介: 刘洋(1987-),男,天津人,北京工业大学硕士研究生,研究方向:计算机仿真,Kinect技术。

Method of Human Dynamic Modeling Based on RGB-D Sequences

  1. Beijing University of Technology, Beijing 100124, China
  • Received:2014-03-24 Online:2014-06-13 Published:2014-06-25

摘要:

人体建模是计算机视觉研究领域的重要研究课题。人体建模被广泛应用于科研、动画、游戏、服装设计、工业等领域,具有非常广阔的应用前景。传统的建模方法可以在大体上还原人体的
姿态,但细节上会有偏差。本文提出一种基于RGB-D序列的人体动态建模方法。人体在场景中自然活动,利用廉价的深度摄像设备Kinect可以获取人体的骨架信息和三维点云。利用获得的骨架信息将模板
人体分段刚性地变形到目标位置,使用ICP算法将变形后的模型与Kinect获取的点云进行更精确的配准,使用TPS变形获得一个平滑的柔性形变人体。

关键词: 人体建模, 深度摄像机, 骨架

Abstract:

Human modeling is an important research topic in the field of computer vision research. Human modeling has a very broad application prospects in the field of
scientific research, animation, games, costume design, industry, etc. The traditional modeling methods can restore the body’s posture in general, but there will be deviations
details. This paper presents a method based on RGB-D sequences. When human activities in the scene, we could capture the data of the human skeleton and three-dimensional point
cloud by cheap equipment Kinect. First, we transform the template human modeling to the target location by the skeleton data captured. Next, the ICP algorithm is adopted to make
the model point cloud acquired by Kinect more accurate registration. Finally, we obtain TPS to make the human modeling smooth deformation.

Key words:  human modeling, depth camera, skeleton