Identification of Wheat Kernels damaged by the Red Flour Beetle using X-ray Images

There is a need for an objective and efficient method to detect insect infestations in the incoming grain in commercial grain-handling facilities. The potential of a soft X-ray method to detect infestations by monitoring feeding damage to kernels caused by the red flour beetle, Tribolium castaneum (Herbst), in wheat was determined in this study. Wheat kernels, artificially infested by T. castaneum eggs were X-rayed continuously for 16 days at 4 day interval for infestations by four larval stages. Algorithms were developed to extract a total of 57 features using histogram groups, textural features, and histogram and shape moments from the X-ray images of wheat kernels. Identification of uninfested and infested wheat kernels was determined using the DISCRIM procedure of the statistical classifiers and a four-layer back propagation neural network (BPNN). Uninfested and kernels infested by four larval instars were correctly identified with more than 73 and 86% classification accuracies by the statistical classifiers and BPNN, respectively, using 57 features. There was no significant difference between the statistical and neural network classifiers for the identification of uninfested and wheat kernels infested by T. castaneum.