Content based texture image retrieval based on modified dominant directional local binary pattern

Abstract

In this paper, a variation in the Local Binary Pattern (LBP) called Modified Dominant Directional LBP (MDDLBP) is proposed. In this method, the direction of the feature with respect to its central pixel in the LBP is preserved by comparing the neighborhood of the pixel in the four dominant directions such as horizontal, vertical, diagonal and anti diagonal. This method captures complete structure information of the texture in the image and also the dimensionality of the feature vector is minimized. The performance of the proposed method is evaluated with two texture datasets namely (i) Brodatz and (ii) VisTex and is compared with state of the art methods using Canberra distance. The proposed method yields better results when compared with other methods.