The automatic detection of hiding narcotics in human bodies based on fractal dimension and SVM classification

In consideration of the harm to society, hiding narcotics in human bodies should be investigate strictly. While the automatic detection method is absent nowadays, and the inspection rate by human eyes is low. So we introduce a new method based on directional fractal dimension texture feature extraction and support vector machine(SVM) to classify the inspection x-ray images. Using this method, the detection could be totally automatic. It has been compared by cross validation in RBF and linear kernel. In our method, the classification accuracy is 95% and negative predictive value is 100% when 40 images had been tested.