Authors:
Asaf Shupo
;
Bill Kapralos
and
Miguel Vargas Martin
Affiliation:
University of Ontario Institute of Technology, Canada
Keyword(s):
Skin detection, surveillance, omni-directional video sensing, maximum likelihood estimator.
Related
Ontology
Subjects/Areas/Topics:
Computer Vision, Visualization and Computer Graphics
;
Image and Video Analysis
;
Image Formation and Preprocessing
;
Segmentation and Grouping
;
Spatial Color Indexing
Abstract:
This paper describes the development of a simple, video-based system capable of efficiently detecting human skin in images captured with an omni-directional video sensor. The video sensor is used to provide a view of the entire visual hemisphere thereby providing multiple dynamic views of a scene. Color models of both skin and non-skin were constructed with images obtained with the omni-directional video sensor. Using a stochastic weak estimator coupled with a linear classifier, the system is capable of distinguishing omni-directional images that contain human skin from those that do not. Results indicate that the system is able to accomplish this task in a simple and computationally efficient manner. The ability to obtain an image of the entire scene from a single viewpoint using the omni-directional video sensor and determine whether the image contains human skin (e.g., one or more humans) in a simple and efficient manner is practical as a precursor for a number of applications inc
luding teleconferencing, remote learning, and video surveillance, the application of interest in this work.
(More)