## 基于RGB-D图像的平面提取算法研究

Kinect传感器获取的有组织的点云。通过在图像空间中均匀地分割这种点云进入点不重叠的组，首先构造曲线图，其节点和边分别代表一组点及其附近的点

Title   Plane Extraction Algorithm Research Based on RGB-D Image
Abstract
Real-time plane extraction in 3D point clouds is crucial to many robotics applications. We achieve a novel algorithm for reliably detecting multiple planes  in organized point clouds obtained from devices such as Kinect sensors. By uniformly piding such a point cloud into non-overlapping groups of points in the image space, it first construct a graph whose node and edge represent a group of points and their neighborhood respectively. We then perform an agglomerative hierarchical clustering on this graph to systematically merge nodes belonging to the same plane until the plane fitting mean squared error exceeds a threshold. Finally we refine the extracted planes using pixel-wise region growing. The experiments demonstrate that the algorithm can reliably achieve 640×480 point cloud plane extraction.

Keywords  point cloud，plane extraction，graph initialization，agglomerative hierarchical clustering

1  引言    1
1.1  研究背景与意义    1
1.2  平面提取算法研究现状    1
1.2.1 现有的平面提取算法    1
1.2.2 现有方法存在的问题    2
1.3  本文的安排    3
2  平面提取的相关知识    4
2.1  数据采集    4
2.1.1  kinect简介    4
2.1.2  数据的特点    5
2.2  平面提取的相关公式    5
3  凝聚层次聚类算法原理    7
3.1  算法概述    9
3.1.1  行回归线段提取算法    12
3.1.2  线段提取与平面提取的差异    13
3.1.3  算法的应用范围    14
3.2  凝聚层次聚类算法构思    14
3.2.1  数据初始化    14
3.2.2  凝聚层次聚类    16
3.2.3  实施细则    17
4  凝聚层次聚类算法实现    19
4.1  程序流程    19
4.2  实验的结果与分析    21

1  引言
1.1  研究背景与意义

------分隔线----------------------------