Title :
Partial image matching by measures from connected color regions
Author :
Yong Kim, Tae ; Han, Joon H.
Author_Institution :
Graduate Sch. of Adv. Imaging Sci. Multimedia & Film, Chungang Univ., Seoul, South Korea
Abstract :
Color information is widely used in content-based image indexing and retrieval, but the information is global, which means that it is computed over an entire image. Thus it is not proper to use it in image indexing where local information is needed. As a method of providing local information, we introduce measures computed from connected color regions. In a connected color region, the color density and maximum co-occurrence color probability are calculated. The measures are used in retrieving an image that contains a partial image which is given as a query
Keywords :
content-based retrieval; database indexing; image colour analysis; image matching; probability; visual databases; color density; color information; connected color regions; content-based image indexing; content-based image retrieval; local information; maximum co-occurrence color probability; partial image matching; Computer science; Content based retrieval; Histograms; Image matching; Image retrieval; Indexing; Information retrieval; Marine vehicles; Probability; Robustness;
Conference_Titel :
Multimedia and Expo, 2000. ICME 2000. 2000 IEEE International Conference on
Conference_Location :
New York, NY
Print_ISBN :
0-7803-6536-4
DOI :
10.1109/ICME.2000.869617