Title: Biped Cartoon Retrieval Using LBG-Algorithm Based State Vector Quantization
Authors: Pakdee, Siriprapa
Kanongchaiyos, Pizzanu
Citation: WSCG '2007: Short Communications Proceedings: The 15th International Conference in Central Europe on Computer Graphics, Visualization and Computer Vision 2007 in co-operation with EUROGRAPHICS: University of West Bohemia, Plzen, Czech Republic, January 29 – February 1, 2007, p. 257-264.
Issue Date: 2007
Publisher: Václav Skala - UNION Agency
Document type: konferenční příspěvek
URI: http://wscg.zcu.cz/wscg2007/Papers_2007/short/!WSCG2007_Short_Proceedings_Final-Part_2.zip
ISBN: 978-80-86943-02-2
Keywords: počítačová animace;kvantizace vektorů;vektor stavu;LBG algoritmus;kreslené filmy;vyhledávání pohybu
Keywords in different language: computer animation;vector quantization;state vector;LBG algorithm;cartoons;motion retrieval
Abstract: Efficiency of cartoon motion retrieval depends on indexing method. Keyword based indexing methods for restoring cartoon motion is simple, but sometimes the process of addressing index to the desired motion is not trivial. Although the methods based on clustering clips of motions or keyframes can be used for retrieval automatically, the complexity of time and reliability depend on size of the database. This research proposes a method for biped cartoon motion retrieval using state vectors of keyframes as the motion indexes. The state vectors derived from the input motion is quantized by our enhanced LBG Algorithm to compute the proper number of indexes for the set of input keyframes. The results show that not only the precision and recall of the cartoon motion retrieval is increased accurately but the process for restoring and retrieval is also performed automatically.
Rights: © Václav Skala - UNION Agency
Appears in Collections:WSCG '2007: Short Communications Proceedings

Files in This Item:
File Description SizeFormat 
Pakdee.pdfPlný text399,12 kBAdobe PDFView/Open

Please use this identifier to cite or link to this item: http://hdl.handle.net/11025/11189

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.