Generalized 3D Co-occurrence Matrix for Visual Information Retrieval *
Xuelong Li, Zhengkai Liu, Nenghai Yu, Yuan Yuan, Dapeng Tao, Wen Yonggang, Xiaoou Tang · Journal of Circuits and Systems · 2002
A novel feature set based on generalized co-occurrence matrix is proposed for Content-based Image Retrieval (CBIR). Texture features provide important cue for the visual information retrieval and discrimination of image content, therefore a set of 3D co-occurrence matrix based texture features is proposed. The preliminary results based on the novel features are much better than traditional ones. Experiments are implemented on an image database including 25000 images.