Abstract :
In this paper, we propose a hybrid SVM-QPSO model based ceramic tube surface defect detection algorithm. The framework of the automatic ceramic tube surface defect detection system is described. In the proposed, the control systems are the key component, which controls electromechanical systems, automatic transfer apparatus and ceramic tube images collecting. The main work of this paper lies in that we classify the ceramic tube images by the hybrid SVM-QPSO model to detect defects on the surface. Afterwards, the parameters of SVM classifier are calculated by Quantum behaved particle swarm optimization, and then we can determine if there is a surface defect on the ceramic tube. Furthermore, the algorithm to compute parameters for SVM based on QPSO is illustrated in detail. To make performance evaluation, experiments are conducted to testify the effectiveness of our method. In this experiment, five kinds of ceramic tube surface defects are considered, and we collect an image dataset which consists of 500 images with ceramic tube surface defects. Comparing with other methods, we can conclude that the proposed algorithm can effectively defects on the surface of ceramic tubes.
Keywords :
automatic optical inspection; ceramics; image classification; particle swarm optimisation; pipes; quantum computing; support vector machines; SVM classifier; automatic ceramic tube surface defect detection system; automatic transfer apparatus; ceramic tube image classification; ceramic tube image collection; control systems; electromechanical system control; hybrid SVM-QPSO model; image dataset; performance evaluation; quantum behaved particle swarm optimization; Ceramics; Electron tubes; Particle swarm optimization; Support vector machines; Surface morphology; Surface texture; Surface treatment; Ceramic tube; Confusion matrix; Parameter selection; QPSO; SVM; Surface defect;
