Search

One of the challenges that need to be addressed to automate grain inspection is the machine detection of insect infestations in grains. In this study, the soft X-ray and NIR spectroscopy methods to detect insect infestations were evaluated for their potential for real-time application. Infested wheat kernels were prepared by artificially infesting Canada Western Red Spring wheat by Sitophilus oryzae adults. Single kernels of wheat uninfested and infested by larvae, pupae, and adults of S. oryzae were first scanned by X-rays. The same infested kernels were then mixed with uninfested bulk grain and scanned using a spectrophotometer. The infestation level in the soft X-ray method was quantified by counting the number of infested and unifested kernels whereas in the NIR spectroscopy method it was quantified by the mass of infested and unifested grain. The identification of infestations by both methods increased with the increase in the developmental stage of the insect from larvae to adult stage. The soft X-ray method has the advantage of potential application in grain inspection over NIR spectroscopy where the number of infested or insect-damaged kernels is an essential information. The NIR spectroscopy analyzing bulk samples has applications in grain management such as fumigation where the identification of insect species is critical and precise quantification of infestations is not vital.

Airflow resistance in grain bulks is a critical factor for the design of drying and aeration systems. Airflow resistance is required to predict uniformity of airflow during aeration and to determine duct designs for adequate ventilation. Previous studies have determined that the airflow resistance in the vertical direction is 30 to 40% higher than in the horizontal direction in cereal grain bulks. This may be caused by the tendency of grain kernels to lie with their major axes being nearly horizontal during filling thus creating narrower air flow paths in the vertical direction than in the horizontal direction. In this study, airflow paths along the vertical and horizontal directions in wheat samples were determined using Xray computer tomographic (CT) scan images. The CT scan images were analyzed using image processing algorithms and air flow paths were estimated. Images were also analyzed to determine the orientation, length and area of airflow paths. It was determined that the area of airflow path along the horizontal direction was about 9% less than in the vertical direction. This difference in airflow path might partially contribute to the difference in airflow resistance along vertical and horizontal directions in grain bulks.

Using soft X-ray it is possible to detect insect infestations in grain samples. Numerous studies have determined the potential of the soft x-ray method to identify insect-infested grain kernels by different types of insects. Only a few attempts have been made to determine the insect types and their developmental stages from infested grain samples that would help to decide management strategies. Mass loss in infested grain kernels depends on the insect type and its life stages. Hence, a correlation between the features obtained from the x-ray images of grain kernels and their mass may be an improved way of determining insect species and their developmental stages. In this study, area and total gray values extracted from the x-ray images of the kernels were correlated to their mass. Single kernels of Canada Western Red Spring wheat was scanned at 15.5 kV and 70 µA with kernels’ crease facing down and up or sideways. For kernels scanned with the crease down, correlation coefficients of over 0.70 were obtained for area and total gray values. However, for kernels with crease up or on the side a poor correlation (r2 = 0.35) was determined between total gray value and mass. This was due to the fact that real time x-ray images have less gray values for dense areas. Hence, all images were transformed into negative images and then analyzed. These images showed a good correlation (r2 above 0.70) between the features from the x-ray images and kernels’ mass for both crease down and up kernels. This relationship may also be used to identify foreign materials and mechanically damaged kernels in grain samples during identification of insect-infested grain kernels.

Large quantities of bulk grain are moved using railcars in Canada and other parts of the world. Automation has not progressed significantly in the grain industry probably because the market is limited for automated systems. A concept of a robot (“Grainobot“) using machine vision to automatically open and close railcar hopper gates is presented. A prototype is being built for further testing and evaluating in a grain handling facility.

An ellipse fitting algorithm was developed and tested for the separation of touching grain kernels in images. This algorithm randomly tracks the edge of touching clusters to find the base points for fitted ellipses. When fitted ellipses were created, a grouping and classifying model was used to identify the best representative ellipse for each kernel of the touching cluster. With representative ellipses, touching grain kernels were separated by morphology transform. Random touching kernel patterns of durum wheat, Canada West Red Spring (CWRS) wheat, oats, barley were used to test this algorithm, the accuracy of separation were: 91%(barley), 98%(durum wheat), 94.5%(oats), and 95%(CWRS wheat).

The efficiency of a soft X-ray method (65 kV and 15 µA) to detect internal seed infestations by the rice weevil, Sitophilus oryzae in Canada Western Red Spring (CWRS) wheat was determined in this study. The infested kernels were identified by the presence of egg plugs and X-rayed every 5 to 7 d until the adults emerged from the kernels. A total of 57 features using histogram groups, histogram and shape moments, and textural features were extracted from the X-ray images. A fourlayer back propagation neural network correctly identified 99% of sound kernels and classed 1% as infested. About 97% of wheat kernels infested by larvae and all kernels infested by pupa and adult stages were correctly identified as infested.