Determination of Sorghum Grain Shape using Image Processing

Abstract

Sorghum grain sphericity, aspect ratio and shape factor were determined from dimensional measurements by image processing and compared to measurements made manually for three sorghum varieties grown in Botswana (Segaolane, Phofu and Botswana Sorghum Hybrid-1). 49 grains per variety were individually scanned using a desktop scanner, and the images processed and analysed using a programme written in Visual Basic 2010TM. Between the two methods, no statistically significant difference (p<0.05) was found for any of the shape descriptors analysed. Variety-wise, however, significant differences (p<0.05) were found using both methods. Across the varieties, the aspect ratio by manual measurement ranged between 0.83 and 0.92, while it ranged between 0.73 and 0.83 when determined using image processing. For sphericity, manual measurement produced a range between 78.11 and 86.26% for the three varieties, while measurement by image processing yielded means ranging between 84.35 and 88.76%, Segaolane being the most spherical. Shape factor ranged between 38.45 and 50.51 with significant differences being noted amongst the varieties. It is concluded that the image processing programme developed in this study is suitable for the measurement of sorghum grain shape and that the above varieties of sorghum can be separated based on their shape.