Image retrieval system
ระบบค้นคืนข้อมูลภาพ
Lekha Chaisorn, D. of Computer Sci., F. of Sci., PSU.
E-mail : clekha@ratree.psu.ac.th
Grant : PSU. collaborative research fund
Published : Research Report
Key words : color image retrieval, color image indexing, digital image retrieval,
color histogram
Early works in searching in a vast information space focus on the use of free-text techniques to locate text documents. The same text-based techniques are applied to retrieve images based on their associated text descriptions. This is not very satisfactory, as images' text descriptions tend to be incomplete and inconsistent. Because of this, recent research has focused on "content-based" retrieval techniques. "content-based retrieval permits the users to search directly on the digital patterns of
image content, such as colors, texture and shapes. Color combinations together with free-text is a
way to issue a query for retrieving images. Therefore, our aim is to develop a query-based image retrieval system and to study how well this scheme can apply to our domain, i.e., images of Thailand. Color histogram approach is adopted to index images, free-text approach is employed for indexing documents, and vector space information retrieval model is employed to develop the retrieval part.
The images of Thai culture, arts, and places of interest are used for our image database. The results of the returned images from a set of the test queries are evaluated in term of precision and recall.
BACK