Design of fabric pattern matching algorithm based on support vector machine

TANG Xue-lian，BI Ming-de，SUN Zhi-gang
（Image Processing and Intelligent Control Key Laboratory of Education Ministry of China，
Huazhong University of Science and Technology， Wuhan 430074， China）

Abstract： In order to solve the problems of template matching with fast and accurate performance during the cloth printing process，a matching algorithm based on support vector machine（SVM） was proposed. The Otsu algorithm and contour extraction algorithms were used to acquire the contour curves of all kinds of shapes in the fabric image. The length and area of the contour curves of the template shape were calculated as simple features to build the first level classifier based on the confidence interval，which was used for coarse matching. Then the Fourier descriptors of both temple and non-template contour curves were extracted for the training of support vector machine，and a SVM classifier was built as the second level classifier for accurate matching. The experiment results show that the proposed template matching algorithm is reliable，accurate and suitable for implementation in industrial occasions.
Key words： Fourier descriptor；contour extraction；template match；support vector machine（SVM）