NA-SODINN: a deep learning algorithm for exoplanet image detection based on residual noise regimes.
Carles CanteroOlivier AbsilCarl-Henrik DahlqvistMarc Van DroogenbroeckPublished in: CoRR (2023)
Keyphrases
- image noise
- learning algorithm
- input image
- image structure
- image features
- image content
- random noise
- multiscale
- background noise
- image analysis
- image data
- image regions
- single image
- additive noise
- low signal to noise ratio
- imaging process
- image classification
- high resolution
- image retrieval
- millimeter wave
- low frequency
- image edges
- image segmentation
- bounding box
- detection algorithm
- low snr
- keypoints
- segmentation method
- image representation
- object detection
- noise free image
- geometric distortions
- salt pepper
- pixel intensities
- noisy images
- test images
- denoising algorithm
- bayesian classifier
- noise free
- gaussian noise
- noise level
- partial occlusion
- hough transform
- segmentation algorithm
- super resolution
- text classification
- edge detection