Dichotomous research on near infrared spectrogram of tomato leaf based on SVM algorithm

A method for classfying anti-nematodes tomato and normal tomato based on near infrared (NIR) spectroscopy technology is presented. Forty anti-nematodes tomatoes and forty normal tomatoes were selected, and leaves spectrograms were achieved.The original spectrograms were preprocessed by normalization and wavelet transform. Support vector machine (SVM) model was established.The optimal parameter combination of the model was found by grid search method. The research indicates that the penalty parameter C mainly affects the recognition rate of model, while the kernel parameter γ mainly affects the prediction rate of model. The recognition rate and prediction rate of SVM model are 98% and 100% respectively. SVM model can identify anti-nematodes tomato and normal tomato well.