Design and Experimental Research of Intelligent Inspection and Classification System for Yuba Skin Quality

At present, the surface quality of Yuba skin is determined by sensory methods. In order to realize intelligent classification detection quality, this study designed a system that automatically determines surfaces based on image processing and support vector machine (SVM) approaches. Specifically, uses preprocessing extract grayscale eigenvalues, gray level co-occurrence matrix (GLCM) run length (GLRLM) eigenvalues sample them as input values for grading system. Through model evaluation three models, SVM was selected according results, different kernel functions were used in training. Based Matlab, software developed designed. Intelligent realized through radial basis function (RBF-SVM) model. The best penalty factor (c = 3.50) parameter value (g 0.98) obtained cross-validation. accuracy 95.31% 94.16% training test sets, respectively. RBF-SVM 93.56%, error less than 5% compared with traditional method grading; thus, feasible can be detection.