ELARKHNU browse.title.page

Зараз показуємо 1 - 1 з 1

Approach to Normalizing the Tissue Macroimages Set to Determine Raw Material Composition
(2026) Lianskorunskyi, K.; Molchanova, M.; Mazurets, O.; Tymofiiev, A.; Мазурець, Олександр Вікторович
The paper presents an approach to normalizing textile macroimage datasets for improving the reliability of automated raw material composition determination using computer vision methods. The study highlights that variations in lighting, scale, texture orientation, sharpness, and shooting conditions significantly affect the stability of textile classification systems. The proposed approach includes informative region selection, geometric and photometric normalization, image quality control, and dataset-level harmonization while preserving diagnostically significant texture features. The methodology is aimed at reducing technical variability and improving the comparability of textile images for further feature extraction and classification. The practical value of the approach lies in its applicability to low-cost textile quality control, automated sorting, and recycling systems based on ordinary digital cameras or mobile devices.