Title :
Retrieval of video objects by compressed domain shape features
Author :
Erol, Bema ; Kossentini, Faouzi
Author_Institution :
Dept. of Electr. & Comput. Eng., British Columbia Univ., Vancouver, BC, Canada
Abstract :
In this paper, we propose shape features that are based on MPEG-4 compressed domain shape information. These shape features can be used for efficient retrieval of both MPEG-4 video objects and video object planes. The proposed shape descriptors are invariant of the shape size, translation, and rotation. Because the shape descriptors are computed using the MPEG-4 compressed domain information, extraction of the features require minimal computation time
Keywords :
data compression; feature extraction; image retrieval; video coding; MPEG-4; compressed domain shape features; computation time; feature extraction; rotation; shape descriptors; shape size; translation; video object planes; video objects; Content based retrieval; Data mining; Feature extraction; Image coding; Image retrieval; Information retrieval; MPEG 4 Standard; Shape measurement; Streaming media; Video compression;
Conference_Titel :
Electronics, Circuits and Systems, 2000. ICECS 2000. The 7th IEEE International Conference on
Conference_Location :
Jounieh
Print_ISBN :
0-7803-6542-9
DOI :
10.1109/ICECS.2000.912966
