National Institute of Informatics - Digital Silk Road Project
Digital Archive of Toyo Bunko Rare Books
:: Similarity-Based Image Retrieval ::
27,218 images are indexed for image retrieval. Text Search is also available.
Search from Clustering (Self-Organizing Map)
- The goal is to automatically extract graphic elements, such as photographs, sketches, maps and paintings, contained in rare books.
- You can search images by image similarity. Color histogram is used as the distance measure between images.
- Due to the prematurity of the method, non-graphical elements such as text are also mistakenly extracted. This problem should be solved in future work.
- The similarity-based image search engine is software developed for similarity-based image retrieval functionality in Digital Typhoon project.
Search Images By Text
- This is equivalent to set the search target to "caption" in Text Search.
- Captions include not only figure captions but also table captions.
- For full text search, please use Text Search.
Copyright (C) 2003-2019
National Institute of Informatics
The Toyo Bunko. All Rights Reserved.