A hopfield recurrent neural network trained on natural images performs state-of-the-art image compression.
Christopher J. HillarRam MehtaKilian KoepsellPublished in: ICIP (2014)
Keyphrases
- natural images
- recurrent neural networks
- image compression
- echo state networks
- natural scenes
- higher order
- multiscale
- complex valued
- denoising
- object recognition
- recurrent networks
- feed forward
- input image
- neural network
- wavelet transform
- image denoising
- image patches
- artificial neural networks
- image coder
- subband
- compression scheme
- hidden layer
- sparse coding
- neural model
- multilayer perceptron
- natural image statistics
- scale invariant
- prior models
- image statistics
- compression ratio
- test images
- image processing
- receptive fields
- statistics of natural images
- power spectrum
- natural image patches
- compressed images
- training set
- computer vision
- visual cortex
- neural network model
- genetic algorithm