Submitted to: American Journal of Agricultural Science and Technology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: September 13, 2013
Publication Date: December 28, 2013
Repository URL: http://dx.doi.org/10.7726/ajast.2013.1009
Citation: Pearson, T.C., Wicklow, D.T., Bean, S., Brabec, D.L. 2013. Sorting of fungal-damaged white sorghum. American Journal of Agricultural Science and Technology. 1:93-103. Interpretive Summary: Fungal infected sorghum grains are not suitable to use in food products and have reduced germination rates. While there are commercial sorting systems that can remove sorghum grains having severe discolorations due to fungal damage, many fungal damaged grains cannot be removed by these systems as the fungi may only cause minor discolorations in the form of several tiny spots on the surface of the grain. To address this issue an improved electronic-optical sorting system was developed to identify and remove sorghum kernels that have both large discolorations and several tiny spots. The system uses a color image sensor and digital processor programmed to detect most types of fungal damaged sorghum. Results indicate that nearly 100% of the grains with large discolorations and 90% of the grains with tiny spots can be separated from clean sorghum with only a 10% error rate on the clean sorghum. The throughput of the system is approximately 30 kg/hr. The sorting system can be used to improve the sorghum quality of food products and seed germination rates and might also be used for other grains or pulse crops for which seeds with localized spots need to be removed.
Technical Abstract: A high-speed, color image-based sorting machine was modified to separate white sorghum with symptoms of fungal damage. Most of the sorghum tested was typically white, but over 27% of the bulk contained grains with fungal damage of various degrees, from severe to very slight. Grains with slight fungal damage were characterized as having several tiny black spots randomly spread across the pericarp surface. To identify small dark spots or blemishes, real-time spot detection algorithms were implemented on a field-programmable gate array (FPGA) directly linked to a color image sensor. Concurrently, grains with large amounts of fungal damage were identified using color histogram algorithms. With the FPGA communicating directly with the camera, image analysis speed was maximized by performing many operations in parallel, including inspection of up to four grains at any given time. Sorting tests indicated that after two passes through the sorter, over 90% of the grains with slight fungal damage and nearly 100% of the grains with large amounts of fungal damage were separated from the original bulk. The germination rates of the grains classified by the sorter as having fungal damage were, on average, 43.5% of those that were accepted by the sorter as undamaged. The hardness of the grains accepted by the sorter was also 4% higher after sorting when compared with the original sample and the rejected grains. This sorting system can be used to improve the sorghum quality of food products and seed germination rates and might also be used for other grains or pulse crops for which seeds with localized spots need to be removed.