Title :
Semantic retrieval by spatial relationships
Author :
Cheung, Kent K T ; Ip, Horace H S
Author_Institution :
Dept. of Comput. Sci., City Univ. of Hong Kong, China
Abstract :
A novel spatial representation of image objects is presented for efficient image retrieval based on spatial relationships. Using our new representation, fast indexing and matching of images by spatial relationships can be accomplished. Our approach takes care of the special requirements for content-based image retrieval (CBIR) as compared to existing spatial representation schemes. Experiments demonstrated the feasibility of semantic spatial retrieval using the proposed spatial representation in CBIR.
Keywords :
content-based retrieval; database indexing; image matching; image representation; image retrieval; visual databases; content-based retrieval; image indexing; image matching; image retrieval; semantic matching; spatial image representation; spatial relationships; Application software; Computer science; Content based retrieval; Image reconstruction; Image retrieval; Image segmentation; Indexing; Internet; Knowledge representation; Pattern recognition;
Conference_Titel :
Pattern Recognition, 2002. Proceedings. 16th International Conference on
Print_ISBN :
0-7695-1695-X
DOI :
10.1109/ICPR.2002.1047784