Detection of damaged wheat kernels using an impact acoustic signal processing technique based on Gaussian modelling and an improved extreme learning machine algorithm

Abstract

Wheat kernel damage is a major source of food quality degradation, and long-term feeding on products from damaged wheat kernels will result in malnutrition or even induce diseases. Therefore, detection of damaged wheat kernels is of significant interest. An impact acoustic signal processing technique based on Gaussian modelling and an improved extreme learning machine approach was proposed for detection of insect and sprout-damaged wheat kernels. Discriminant features extracted from Gaussian-model-estimated parameters were fed to an extreme learning machine based on a C-matrix embedded optimisation approximation solution. The best results, 92.0% of undamaged, 96.0% of insect-damaged, and 95.0% of sprout-damaged wheat kernels were correctly classified by using the proposed method. Furthermore, the detection system had good processing speed. Therefore, it could be effective to detect damaged wheat kernels in real time.