Title :
The thread form detection based on SVM
Author :
Yu, Zhong-Dang ; Wang, Long-shan
Author_Institution :
Coll. of Mech. Sci. & Eng., Jilin Univ., Changchun, China
Abstract :
This paper presents a detection method of thread image based on support vector machine. This method is to obtain a training sample set by collecting every concrete sub-image of thread form from the standard thread images as positive examples, and unqualified thread images as negative examples; and to train support vector machine with sequential minimal optimization algorithm, detect the thread with the optimal decision function that was obtained. Experiments show that by this method the detection result is reliable, and it is easy to accomplish the online detection. In addition, this paper also briefly discusses how to reduce point dimension and how to speed up operation.
Keywords :
image recognition; learning (artificial intelligence); minimisation; object detection; support vector machines; SVM; machine learning; optimal decision function; sequential minimal optimization; support vector machine; thread form detection; thread image detection; Concrete; Educational institutions; Information science; Inspection; Machine learning; Machinery production industries; Optimization methods; Pattern recognition; Support vector machines; Yarn; Support vector machine; sample; sequential minimal optimization algorithm; thread;
Conference_Titel :
Machine Learning and Cybernetics, 2005. Proceedings of 2005 International Conference on
Conference_Location :
Guangzhou, China
Print_ISBN :
0-7803-9091-1
DOI :
10.1109/ICMLC.2005.1527710
