A two-step approach to describing web topics via probable keywords and prototype images from background-removed similarities.
Junbiao PangFei TaoLiang LiQingming HuangBaocai YinQi TianPublished in: Neurocomputing (2018)
Keyphrases
- keywords
- web images
- input image
- image database
- high contrast
- web pages
- three dimensional
- image collections
- image registration
- visual similarity
- image retrieval
- image analysis
- image data
- website
- web documents
- similarity measure
- textual information
- complex background
- background information
- image classification
- edge detection
- ground truth
- object recognition
- visual features
- image matching
- information retrieval
- search engine
- semantic categories
- text information
- corel image
- textual descriptions
- global thresholding
- bounding box
- image set
- test images
- semantic web
- web applications
- lighting conditions
- image quality
- image features
- image segmentation