Noise or Signal: The Role of Image Backgrounds in Object Recognition.
Kai Yuanqing XiaoLogan EngstromAndrew IlyasAleksander MadryPublished in: CoRR (2020)
Keyphrases
- random noise
- image features
- object recognition
- template matching
- additive noise
- image noise
- low frequency
- transformed image
- additive gaussian noise
- quantization noise
- image representation
- low snr
- image data
- background noise
- image edges
- computer vision
- power spectrum
- frequency domain
- image content
- high frequency
- image structure
- white noise
- single image
- wiener filter
- image retrieval
- geometric distortions
- noise variance
- high frequency components
- image classification
- bandpass
- low signal to noise ratio
- visual object recognition
- signal noise ratio
- pixel intensities
- high resolution
- input image
- gaussian noise
- test images
- multiscale
- image restoration
- impulse response
- image denoising
- noisy images
- pixel values
- phase information
- noise level
- natural images
- multiplicative noise
- signal detection
- denoising
- noise model
- d objects
- denoising algorithm
- denoising methods