The Use of Support Vector Machine in The Classification of Lachack Toranj and Torkaman Geometric Rug Designs

Design is one of the most critical aspects of evaluating Persian rugs. While Persian rugs are woven in different designs, their unique patterns distinguish them from other rug styles such as Turkish, Indian and Chinese. In this study, the classification of two categories of Persian rugs (Lachak Toranj and Torkaman Geometric) was followed using Support Vector Machine (SVM) and based on Fourier feature. In this regard, the grayscale images of 60 rugs with different designs were collected. Using the Fourier transform, the unique feature of these two rugs’ categories were extracted and used for classification. For this purpose, a two-class support vector machine was used. According to the results, the SVM classification method can classify these two categories of rugs with an accuracy of 89% by using the polar Fourier feature.