Remote sensing imagery retrieval based-on gabor texture feature classification
Hongyu Yao, Bicheng Li, Cao Wen · 2005
According to the remote sensing imagery, a novel method for retrieval based on the classification of Gabor texture features is proposed in this paper. The method firstly applies Gabor transforms with different scales and orientations to the image. Secondly conduct an unsupervised classification of the obtained Gabor features in the way of K-means. At last a set of features based on the classification information are extracted to present the content of the image. Compared with traditional method of Gabor filters, our approach introduces the correlation of the similar textures, omits the effect of unimportant pixels with sparse distribution. Experimental results approve that the novel method do achieve a promotion performance in remote sensing imagery retrieval.