An Efficient Texture Classification Technique Based on Semi Uniform LBP

Abstract

Local Binary Pattern' (LBP) operator has been proved to be a theoretically simple yet very effective statistical texture descriptor in many areas. One of the main disadvantage of LBP is they are more sensitive to even a simple noise, this causes a change in LBP pattern. A LBP can generate 256 (0 to 255) different patterns. These patterns are grouped into 59 uniform and 197 nonuniform LBP patterns. Most of the researchers have considered only uniform patterns for texture classification due to their small numbers and claiming that most of the textures are dominated by only uniform LBP's. But after a careful, critical and significant study on the literature, ULBP have some shortcomings: they discard some important texture information, suffer much from non-monotonic illumination variation and do not describe the stochastic characteristics of texture efficiently and sensitive to noise. To overcome this, the present paper defined 'Semi Uniform LBP' (SULBP). The proposed