Rotation-invariant image retrieval algorithm based on texture
Tao Zhang · Jisuanji gongcheng yu sheji · 2005
A rotation-invariant and multi-scale generalized roughness feature is presented and itwas combinded with an adaptive weighted distance to apply in texture retrieval.Texture image was decomposed based on Gabor transformation,the rotation-invariant vector was formed by making use of the amplitude spectrum and combining with energy distributions,Hurst parameters and direction differences.Finally,the adaptive weighted distance was used to measure the similarity.Simulationresults show that the proposedapproach performed well in the rotation texture images.