Detection and segmentation of image anomalies based on unsupervised defect reparation.
Wenting DaiMarius ErdtAlexei SourinPublished in: Vis. Comput. (2021)
Keyphrases
- segmentation method
- image segmentation
- segmentation algorithm
- image regions
- image analysis
- grey level
- circular hough transform
- image features
- pixel level
- input image
- test images
- edge detection
- intensity variations
- multiscale
- anomaly detection
- bounding box
- image pixels
- color segmentation
- foreground background separation
- single image
- image content
- segmentation accuracy
- image segmentation algorithm
- image data
- multiple objects
- gray level images
- complex background
- gray value
- segmented images
- hough transform
- energy function
- energy functional
- perceptual grouping
- image retrieval
- homogeneous regions
- region segmentation
- semantic segmentation
- adaptive thresholding
- texture and shape features
- automatically segmented
- fundus images
- segmentation errors
- segmented regions
- low signal to noise ratio
- object segmentation
- contour extraction
- gray level
- optimal segmentation
- intensity distribution
- image structure
- level set
- foreground and background