Click on image for animation (1.0 MB)
We use chromakey extraction to check the person we find in the images. This is like television technology, where the background to the subjects is added in. We chose green as this is unlikely to be in many people's clothes (and especially their shoes!).
Click on image for animation (300 KB)
We also extract subjects using new techniques which isolate moving objects. Here, we can see the people moving as well as the tree. We can further process this to find the largest moving blocks, and by more intelligent analysis to find the subjects and recognise them.
Click on image for animation (650 KB)
We also use edge data in model-based feature extraction. Here's some data derived by application of the Canny operator. It might look a little grainy, but that's more the resolution of the display. We use high-level feature extraction operators to group the edge data to find the moving (human) subjects.
Last modified: 04 July 2001 by
M S Nixon
© University of Southampton