Multirate Kalman filter rejects impulse noise in frequency-domain-multiplexed tracker measurements.
Robert A. MacLachlanRalph L. HollisBranislav JaramazCameron N. RiviereJoseph N. MartelKenneth L. UrishPublished in: IEEE SENSORS (2017)
Keyphrases
- frequency domain
- kalman filter
- impulse noise
- particle filter
- object tracking
- filter bank
- subband
- mean shift
- median filter
- color images
- gaussian noise
- denoising
- spatial domain
- kalman filtering
- noise removal
- noise level
- noise reduction
- digital images
- visual tracking
- noise free
- appearance model
- feature vectors
- image sequences
- computer vision
- feature extraction
- motion detection
- fourier domain
- image processing
- multiresolution
- wavelet transform
- filtering method
- video sequences
- motion model
- noisy images
- gray scale
- machine learning
- data association
- high quality
- state space
- image restoration
- natural images
- wavelet coefficients
- moving objects
- pattern recognition
- computational complexity
- feature space