Texture Classification Using Dominant Gradient Descriptor

Texture classification is an important part of manyobject recognition algorithms. In this paper, a new approach totexture classification is proposed. Recently, local binary pattern(LBP) has been widely used in texture classification. Inconventional LBP, directional statistical features and colorinformation are not considered. To extract color information oftextures, we have used color LBP. Also, to consider directionalstatistical features, we proposed the concept of histogram ofdominant gradient (HoDG). In HoDG, the image is divided intoblocks. Then the dominant gradient orientation of each block ofimage is extracted. Histogram of dominant gradients of blocks isused to describe edges and orientations of the texture image. Bycoupling the color LBP with HoDG, a new rotation invarianttexture classification method is presented. Experimental resultson the CUReT database show that our proposed method issuperior to comparable algorithms.

Keywords:

texture classification , local binary pattern histogram of dominant gradient , rotation invariance