Landmine recognition research based on SVM

Support Vector Machine (SVM) is superior in resolving the recognition problem in which the sample is less and the dimension of the data space is high. In order to realize the landmine auto recognition, in this paper, the landmine classification and recognition are realized using SVM in conjunction with genetic algorithm. But, the signal of the object landmine is often deteriorated by strong clutter; in order to reduce the clutter and improve the performance of recognition, a method based on wavelet packet is used in the field of radar data preprocessing. Experiment results prove that the method of feature selection and classification recognition is valid; this method can resolve the problem of high dimension data and less sample, the recognition ratio can reach 86%.