A Texture Descriptor Combining Fractal and LBP Complex Networks

There is a growing interest in multilabel image classification. In this study, we proposed a novel texture classification approach combining the fractal theory and the LBP complex networks (FLCN). The complex networks were constructed based on the LBP features and the pixel relationships of the image. The suitable parameters and the combination of statistical properties on the complex networks were investigated to represent the texture of the image. The experimental results show that the FLCN method has good performance in the classifications of the segmented images and the biomedical images. In addition, the approach is more robust compared with other methods.