Application Overview

This application was written specifically for a Silicon Graphics Onyx, but with some effort, could be ported to another platform. Hardware specifics currently limit the portability. It supports single images, as well as video from a sequence of files, and video from a Sirius Video frame grabber board. The application was designed around the idea that a single person would be present in a video image, and this person would use his right hand make some sort of hand gesture (see Figure 4 for an example). It was also assumed this person would be wearing a non-skin tone long sleeve shirt when in the image. There are many command-line switches available including "-h" for help. The display interface displays the input image and the resulting output image after segmentation. Keyboard commands can be used to invoke color predicate training if desired.