A Novel Texture Descriptor Evaluation Window Based Adjacent Distance Local Binary Pattern (EADLBP) for Image Classification

Abstract

In this research, we suggested a novel texture descriptor distance-based Adjacent Local Binary Pattern AdLBP based on the adjacent neighbor window and the relationships among the sequential neighbors pixel value with a given distance parameter. The suggested technique calculates the neighbor and extracts the binary code from the adjacent neighborhood window and surrounding sub-image window in order to improve the adjacent neighbor information and change the conventional LBP thresholding schema. Additionally, we expanded this adjacent distance-based local binary pattern AdLBP and combined it with the evaluation window-based local binary pattern EwLBP to create a texture descriptor for texture classification that is more robust texture descriptor against noise. Finally combine AdLBP And EwLBP using encoding strategy to propose an Evaluation window based on Adjacent Distance Local Binary Pattern EADLBP descriptor for Image Classification. These descriptors are tested with the KTH-TIPS, KTH-TIPS2b to the applicability of the proposed method. In comparison, the proposed EADLBP approach is more robust against noise and consistently out- perform all of the fundamental methods.