Performance Evaluation of Different Local Binary Operators for Texture Classification

• Main Authors: Shamaileh, Abeer, Rassem, Taha H., Liew, Siau-Chuin, Al Sayaydeh, Osama Nayel
• Format: Article
• Language: English
• Published: The Science and Information (SAI) Organization Limited 2019
• Subjects: QA76 Computer software
• Online Access: http://umpir.ump.edu.my/id/eprint/24773/1/Performance%20Evaluation%20of%20Different%20Local%20Binary%20Operators%20%20for%20Texture%20Classification.pdf; http://umpir.ump.edu.my/id/eprint/24773/; https://thesai.org/Publications/IJACSA
• Institution: Universiti Malaysia Pahang

Local Binary Patterns (LBP) have brightened up as one of the most eminent and widely studied texture descriptors. The LBP descriptor has gained high acceptance due to its simplicity, high distinguishing power, and flexibility. As such, it has been deployed in several applications where it has performed well. This is why LBP is the basis for a new research direction. Several forms of LBP have been suggested to increase its discriminative ability during texture classification and to improve its robustness to noise. Since 2002, different texture descriptors had been proposed. These texture descriptors were inspired by LBP and proposed to overcome its limitations. Examples of these texture descriptors are Local Ternary Pattern (LTP), Completed Local Binary Pattern (CLBP), Completed Local Binary Count (CLBC), Completed Local Ternary Pattern (CLTP), and Wavelet Completed Local Ternary Pattern (WCLTP). Due to the importance of texture descriptors in image classification, the performance of different texture descriptors is studied and investigated for image texture classification in this paper. This study also strived to improve the role of image texture information in classification processes. Different experiments were conducted using two benchmark texture datasets - CuRTex and OuTex. The experimental results showed that the WCLTP outperformed the remaining texture descriptors. The WCLTP achieved 99.35% and 96.89% classification performance accuracy with CuRTex and OuTRex, respectively.