Multiscale Feature-Clustering-Based Fully Convolutional Autoencoder for Fast Accurate Visual Inspection of Texture Surface Defects.
Hua YangYifan ChenKaiyou SongZhouping YinPublished in: IEEE Trans Autom. Sci. Eng. (2019)
Keyphrases
- visual inspection
- textural features
- multiscale
- printed circuit boards
- surface defects
- image analysis
- laser scanning
- color texture
- fluorescence microscopy images
- registration accuracy
- image representation
- coarse to fine
- image processing
- machine vision
- texture analysis
- restricted boltzmann machine
- wavelet transform
- feature vectors
- texture features
- texture classification
- texture segmentation
- conditional random fields
- natural images
- edge detection
- image features
- fine scale
- high quality
- feature set
- image segmentation