Supervised classification and segmentation of textured scene images

Texture classification of the textured scene images is a challenging task due to variation of textures in images. In addition, occlusion, reflection and shadows make texture classification more difficult. We propose a novel patch-based approach for texture segmentation and classification in the textured scene images. First, color and texture features are designed to extract crucial information from non-overlapping patches. Each patch is classified into one of the predefined categories by the random forest classifier. Then the labeled result is refined with the assistance of the conventional image segmentation method. The experimental results show that the proposed method outperforms current state-of-the-art approaches in terms of speed and accuracy.