Texture classification approach based on combination of random threshold vector technique and co-occurrence matrixes

Texture classification became one of the problems which has been paid much attention on by image processing scientists since late 80s. Consequently, since now many different methods have been proposed to solve this problem. In most of these methods the researchers attempted to describe feature´s set which provide good dimensionality and severability between textures. In RTV method, a new feature´s set derived from the fractal geometry is called the random threshold vector (RTV) for texture analysis. The results have shown, this method can´t provide high accuracy rate in texture classification. So in this paper an approach is proposed based on combination of RTV and Co-occurrence matrixes. First of all, by using a unique threshold method the first dimension of feature vector is calculated. After that, by using RTV method, the entropy is computed of Co-Occurrence matrixes. So, the vectors have two dimensions, one of them is threshold dimension and another is the entropy´s value for the co-occurrence matrix. In the result part, the proposed approach is applied on some various datasets such as Brodatz and Outex and texture classification is done. High accuracy rate shows the quality of proposed approach to classification textures. In addition the random threshold vector technique based on co-occurrence matrix contains great discriminatory information which is needed for a successful analyzed. This approach can use in various related cases such as texture segmentation and defect detection.