Detecting sugarcane borer diseases using support vector machine

Based on the fact that great labor of artificial selection was needed after the sugarcane seeds were cut by the sugarcane cutting machine, and there was a misjudgment of the sugarcane borer diseases. SVM (support vector machine) method was proposed in this study to detect the sugarcane borer diseases. With the machine vision technology, together with threshold segmentation, filling and corrosion operation to process the three images of the same sugarcane whose interval is 120°. The classification features, minimum average gray value and the corresponding minimum gray value were selected by adaptive threshold segmentation algorithm, and removed the region which area of 1. The study used radial basis function as the kernel function of SVM, and roughly selected the range of regularization parameters of C and kernel function parameter σ. Finally, it selected the optimal parameters by the grid search and the cross validation method to identify sugarcane with diseases. The test showed that correct rate of diseases and disease-free sugarcane is 96%, 95.83% for the test set, so the method can effectively complete the sugarcane borer diseases detection.