Title :
A novel multi modal tracking method based on depth and semantic color features for human robot interaction
Author :
Chandarr, Aswin ; Rudinac, Maja ; Jonker, Pieter
Author_Institution :
Fac. of Mech. Eng, Delft Univ. of Technol., Delft, Netherlands
Abstract :
In this paper we tackle the challenges of visual tracking for personal robots. We have proposed a novel track-by-detection method that combines a semantic object model with depth properties to obtain target contours. The tracking can be initialized by either 2D or 3D inputs, which are further refined using clustering based background removal to obtain an initial object model. During tracking, we propose to refine the search space by metric constancy and removal of the support plane. Further, the target appearance is modeled using a semantic human centric color descriptor, continuously updated by an online learning algorithm. The spatial compactness of the target is described using a Gaussian model with an initially determined variance. A fusion of the obtained color and depth models based on a target-background dissimilarity measure, is used to perform segmentation based tracking using graph cuts to obtain object contours. The experimental results in a household scenario show a good performance of the algorithm in challenging conditions such as scale, viewpoint change and out of plane rotations.
Keywords :
Gaussian processes; edge detection; feature extraction; graph theory; human-robot interaction; image colour analysis; image segmentation; learning (artificial intelligence); object tracking; pattern clustering; robot vision; search problems; Gaussian model; clustering based background removal; depth features; depth properties; graph cuts; human robot interaction; initial object model; metric constancy; multi modal tracking method; online learning algorithm; out-of-plane rotations; personal robots; scale; search space; segmentation based tracking; semantic color features; semantic human centric color descriptor; semantic object model; spatial compactness; support plane removal; target appearance modeling; target contours; target-background dissimilarity measure; track-by-detection method; viewpoint change; visual tracking; Color; Human-robot interaction; Image color analysis; Object recognition; Robots; Semantics; Target tracking;
Conference_Titel :
Machine Vision Applications (MVA), 2015 14th IAPR International Conference on
Conference_Location :
Tokyo
DOI :
10.1109/MVA.2015.7153207