Title
Autonomous Object Segmentation In Cluttered Environment Through Interactive Perception
Abstract
This paper investigates the problem of object segmentation in cluttered environment. This problem enables a large variety of exciting and important applications. An interactive perception method is proposed to segment scene into constituent objects based on principal angle. Trajectory data of feature points reflecting the essence of scene structure changes is extracted by a robot arm to calculate least stable regions for the interactive task. The segmentation task is achieved by the principal angle of stable regions because the principal angle essentially estimates the similarity between two regions. In contrast to probability based approach, our method performs well on efficiency of segmentation as it works without estimating parameters of the system. Experimental results on real world scene confirm the effectiveness of our method.
Year
DOI
Venue
2017
10.1007/978-3-319-67777-4_30
INTELLIGENCE SCIENCE AND BIG DATA ENGINEERING, ISCIDE 2017
Keywords
Field
DocType
Object segmentation, Interactive perception, Trajectory, Principal angle
Computer vision,Robotic arm,Principal angles,Computer science,Segmentation,Artificial intelligence,Perception,Trajectory
Conference
Volume
ISSN
Citations 
10559
0302-9743
0
PageRank 
References 
Authors
0.34
9
5
Name
Order
Citations
PageRank
Rui Wu195.26
Dongfang Zhao236226.49
Jiafeng Liu314018.43
Xianglong Tang428844.84
Qingcheng Huang500.34