Innovative pattern based morphological method for texture segmentation

Texture segmentation has been active area of research for over three decades. Texture segmentation is the method of dividing an image into homogenous regions. Morphology based approaches to texture segmentation have gained popularity in recent years in the general computer vision literature because of their capability to give a globally optimal solution. In this paper, a new morphological approach based on local tetra patterns (LTrPs) is proposed for the segmentation of textures. The standard local binary pattern (LBP) and local ternary pattern (LTP) encode the relationship between referenced pixel and its neighbors by computing gray-level difference whereas LTrP encodes the relationship between the referenced pixel and its neighbors, based on the directions that are calculated using the first-order derivatives in horizontal and vertical directions. The algorithm is tested on Brodatz and VisTex databases and the results obtained show good segmentation.