Hardware implementation of the Gaussian Mixture Model foreground object segmentation algorithm working with ultra-high resolution video stream in real-time.
Piotr JanusTomasz KryjakPublished in: SPA (2018)
Keyphrases
- segmentation algorithm
- hardware implementation
- video streams
- gaussian mixture model
- foreground objects
- object segmentation
- background subtraction
- background model
- high resolution
- video data
- video frames
- superpixels
- mixture model
- image segmentation
- video sequences
- signal processing
- segmentation method
- foreground and background
- feature vectors
- graph cuts
- pixel level
- maximum likelihood
- video surveillance
- expectation maximization
- input image
- em algorithm
- dynamic scenes
- moving camera
- field of view
- shape prior
- image processing
- super resolution
- object detection
- motion detection
- moving objects
- feature space
- machine learning
- template matching
- multiscale