A new K-View algorithm for texture image classification using rotation-invariant feature
Hong Liu, Siguang Dai, Enmin Song, Cihui Yang, Chih‐Cheng Hung · 2009
This paper proposes a new K-View algorithm for texture image classification using rotation-invariant features. These features are statistically derived from characteristic view sets for each texture. Unlike the existing K-View algorithm, all the views used are transformed into rotation-invariant features and the K views are selected randomly. In contrast, the existing K-View algorithm uses the K-means algorithm for choosing the K views. In this new algorithm the decision of determining a pixel to which texture class it belong to, is made by considering all the views which consist of the pixel being classified. In order to preserve the primitive information of a texture class as much as possible, the proposed algorithm randomly selects k views of the view set from each sample sub-image as the characteristic view set. Experimental results show that the proposed algorithm is more robust and accurate compared with the results of the existing K-View algorithm.