Object Detection Based on Color Features
by
Abhinav Jain
An intelligent system which is capable of detecting human faces should be able to detect human skin. It has been suggested that features which provide an accurate classification between object and a background in an image, should be the best detectors for that object as well. This paper investigates how learning rules based on color features influence object detection capability of an intelligent system. Our results indicate that the color features provide us with a very accurate and a powerful classification between object and non object pixels of an image. The object was skin in most of our experiments.
Advised by Raj Rao
MGH 251
Wednesday
March 3, 2004
3:30 - 4:20 pm