Automatic Estimation of the Legibility of Binarised Historic Documents for Unsupervised Parameter Tuning.
Martin StommelGideon FriederPublished in: ICDAR (2011)
Keyphrases
- parameter tuning
- ink bleed
- document collections
- information retrieval
- parameter settings
- document retrieval
- supervised learning
- web documents
- topic modeling
- text documents
- document clustering
- unsupervised learning
- parameter estimation
- text classification
- machine learning
- document classification
- cultural heritage
- image segmentation
- vector space
- relevant documents
- information retrieval systems
- xml documents
- metadata
- user queries