UDC 004.85
DIRECTIONAL DIFFERENCE VECTOR FOR TEXTURE CLASSIFICATION OF LEATHER SEMI-FINISHED PRODUCTS
А. V. Levitin, PhD (in technical sciences), associate professor, Department of Automation and Information
Technologies in Control, RSREU, Ryazan, Russia;
orcid.org/0000-0002-4829-7398, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
The objective of this study is to evaluate the effectiveness of the 8-point Local Difference Vector
(8-LDV) as a descriptor for texture classification and defect detection in semi-finished leather images. A
comparative analysis was conducted between 8-LDV and traditional methods, including Local Binary Patterns
(LBP), Hessian Eigenvalues (HEV), and statistical features of the Gray-Level Co-occurrence Matrix
(GLCM). Experimental results demonstrate that the use of 8-LDV in combination with Quadratic Discriminant
Analysis (QDA) provides higher accuracy in anomaly classification while significantly reducing computational
costs compared to the combination of traditional descriptors and Linear Discriminant Analysis
(LDA). The proposed 8-LDV + QDA framework is recommended as a universal and high-performance
method for automated quality control across various leather types.
Key words: texture classification, leather texture, machine learning, texture features, technical vision,
directional difference vector