Surface curvature based completed local ternary pattern for texture image classification

The curvature of two-dimensional function can describe the degree of surface curvature. When an image is treated as a discrete two-dimensional function, image curvature describes the structural relationship between local pixels of the image. Local ternary pattern is an effective image texture descriptor to encode shape index based on image curvature. In this paper, the completed local ternary pattern, which contains the symbol characteristics, amplitude characteristics and central pixel characteristics of the local ternary pattern of shape index (completed local ternary pattern based on shape index, SI-CLTP) are all considered at the same time. Experiments on two texture databases and one palmprint database fully show that shape index based completed local ternary pattern is an effective image texture descriptor.