DETECTION OF INSECTS IN BULK WHEAT SAMPLES WITH MACHINE VISION

Digital imaging techniques were used to identify insects and body parts (adult lesser grain borer) of Rhyzopertha dominica beetles in bulk wheat samples. The samples contained dockage (weed seeds and damaged wheat kernels), which made the identification of insects more difficult. Multispectral analysis (Red, Green, Blue) was used in combination with pattern recognition techniques. Insect identification experiments were done with subimages of 8 × 8 pixels. Statistical multivariate analysis and structural patterns were used for finding insects. The position of the insect (ventral, dorsal, side) and particles clinging to the insect affected recognition success. Recognition success of R. Dominica adults, some grass seeds, and other non-wheat components was higher than 90%.