Sparse Coding and Mid-Level Superpixel-Feature for ℓ0-Graph Based Unsupervised Image Segmentation.
Xiaofang WangHuibin LiSimon MasnouLiming ChenPublished in: CAIP (2) (2013)
Keyphrases
- sparse coding
- mid level
- superpixels
- natural images
- image patches
- multiscale
- input image
- feature vectors
- perceptual grouping
- sparse representation
- image classification
- higher order
- image representation
- low level
- low level features
- object segmentation
- object recognition
- unsupervised learning
- denoising
- generative model
- event detection
- high level
- image features
- saliency map
- segmentation algorithm
- semi supervised
- visual information
- test images
- multiple images
- conditional random fields
- high dimensional
- long range
- feature space
- bounding box
- bayesian networks
- graph structure
- active learning
- visual words
- action recognition
- image data
- ground truth
- object detection