Image retrieval based on the texture feature
Zhou Li-hua · Journal of Optoelectronics·laser · 2008
A new image retrieval algorithm based on the texture feature is presented.The orientation-normalized image is achieved by rotating its principal axis to horizontal direction according to the maximum of angular moments.The adoptive wavelet,which has the property of translation and scale invariant,is used to deal with the orientation-normalized image.Then the wavelet coefficients of image have the property of rotation,translation and scale invariance.The wavelet energy of every sub frequency bands is extracted to be the texture feature of the iamge.Gaussian model is used to normalize the different sub-characters distance to the texture feature of image.The texture similarity of the querying image and other images is computed by the Eculidean distance.Experiments indicate that this method has the property of rotation,translation and scale invariance in image's similarity retrieval and has a higher retrieval-rate than other methods.