Universal colour quantisation for different colour spaces
Chia-Mu Yu, Songyan Chen · IEE Proceedings - Vision Image and Signal Processing · 2006
A universal colour quantisation for different colour spaces is proposed. First, the differences between colour spaces such as RGB, YCrCb, YIQ, I1I2I3, HSV, L*a*b* and L*u*v* are analysed. On the basis of analysis results, the relationship between RGB and HSV is derived, which in turn is used to design a universal colour quantisation for improving retrieval effectiveness. Moreover, colour distribution of real images is further considered so as to obtain more accurate retrieval results. The effectiveness and practicability of the proposed method have been demonstrated by various experiments.