Multi-Class Support Vector Machines For Texture Classification Using Gray-Level Histogram And Edge Detection Features
Identification of the machining process producing a specific engineering surface is very important in manufacturing facilities. Computer vision has become center-stage in automatic identification of these processes, with benefits of man-power reduction as well as the drawbacks of human involvement such as inconsistencies caused by fatigue. In this paper, we propose a computer vision framework that takes into consideration workpiece imagesí intensity histogram and edge features to identify the six machining processes of Grinding, Turning, Horizontal Milling, Vertical Milling, Shaping and Lapping. The support vector machine (SVM) classifier is explored with various kernels being investigated. The experimental results show that the SVM with the linear kernel using edge feature statistics provides the best performance for a dataset that consists of seventy-two workpiece images.
Keywords- Machined Texture Classification, Support Vector Machines, Gray-Level Histogram, Edge Detection, PCA