An approach for Fusarium diseased corn kernels recognition using linear discrete models

A new approach to identify infected with pink Fusarium maize seeds through the spectral features in the near infrared region is proposed in the paper. It is based on analysis of coefficients of linear parametric models of discrete type Autoregresion (AR). Seeds identification criterion is based on the boundary of А between the class healthy and class infected seeds. The maximum distance between the two classes – ΔA for the 10th order of AR-model is used to LV determine the boudary. The recognition accuracy achieved 100% for a variety XM87/136 and for varieties 26a, Knezha 436 and Rouse 424 the accuracy range was from 97.50 to 98.75%.

An approach for Fusarium diseased corn kernels recognition using linear discrete models

P. Daskalov, V. Mancheva, T. Draganova, R. Tsonev