An Improved K-view Algorithm for Texture Classification Using Voting Method

Texture is an important characteristic of images. Image texture classification is widely used in many applications, such as detection of industry product, document class, and digital image processing of remote sensing and so on. Several K-View based algorithms have been proposed for image texture classification. This study proposed an improved k-view algorithm for texture classification using voting method. In the first step, the method was implemented in our previous study that all the views used are transformed into rotation-invariant features is employed. Then K views are selected randomly as those characteristic views to form characteristic view sets for each class texture. In the third step, we use those characteristic view sets to class a texture image and achieve a preliminary result. At last, voting method is used to improve the preliminary result. A lot of experiments are carried out using artificial texture images taken from the Brodatz. The experimental results showed that he proposed improved method is more robust and accurate by compared with the old method.