Background subtraction in varying illuminations using an ensemble based on an enlarged feature set.
Brendan KlareSudeep SarkarPublished in: CVPR Workshops (2009)
Keyphrases
- background subtraction
- feature set
- random forest
- base classifiers
- feature selection
- video surveillance
- foreground objects
- object detection
- moving objects
- motion detection
- classification accuracy
- image sequences
- feature vectors
- feature extraction
- background model
- feature space
- gaussian mixture model
- appearance model
- single image
- foreground detection
- background modeling
- moving camera
- texture features
- background image
- dynamic scenes
- ensemble methods
- post processing
- dimensionality reduction
- training data
- model selection
- knn
- extracted features
- three dimensional
- lighting conditions
- learning algorithm