Documentation for Video/DominantColor

Technique Dominant Color
Document MPEG-7 Visual (ISO/IEC 15938-3)
Name Jungmin Song, Heon Jun Kim, Leszek Cieplinski, Prof. Manjunath
EMail jmsong73@mail.lgcit.com, hjk@lge.co.kr, Leszek.Cieplinski@vil.ite.mee.com, manj@ece.ucsb.edu
Contact Jungmin Song(general), Leszek Cieplinski(color variance), Prof. Manjunath(dominant color extraction and search algorithm)
Type application
External Libraries none
Related Ds/DSs Color Space Descriptor, Color Quantization Descriptor, Related DS's are not defined fully yet.
Used Ds/DSs Color Space Descriptor, Color Quantization Descriptor
Input Images
Extraction Yes
Client Appl Search, Retrieval, Browsing
Summary The dominant color descriptor is useful for image and video retrieval. It targets content-based retrieval for color, either for the whole image or for any arbitrary shaped region (rectangular or irregular). It is a very compact descriptor, requiring less than 6-8 colors per region. Since colors are not pre-quantized as in the histogram type color descriptors, the representation is more accurate. To accomplish high accuracy in retrieval, Spatial Coherency and/or Color Variance can be utilized. It is intended for applications that use object based representations (objects or regions in an image).
Limitations The maximum allowed number of dominant colors is 8.
Known Problems None.
Parameters color space and color quantization parameters