Study of commercial grades of Canada Western Red Spring wheat by digital image processing
Kohler, Jurgen Max
Canada's grading system is primarily based on the subjective interpretation of grading factors that influence end-use quality. Grade is determined by grain inspectors upon visual inspection and assessment of sampled grain relative to established standard grade samples. The lack of objective measurement of important grading factors has led to inconsistent grade determinations to the extent that anywhere from 10% to 14% of reinspected samples receive a higher grade from that originally assigned. End-users will always prefer a precise and consistent measurement of quality characteristics, especially as wheat utilization and processing technology is increasing in sophistication. The present study uses digital image processing techniques for the objective measurement of kernel morphology and its uniformity in three commercial grades of Canada Western Red Spring (CWRS) wheat. A total of 32 variables quantified aspects of kernel size, kernel shape and kernel brightness, as well as their relative uniformity within each grade. The analysis is based on 103 carlot and 73 cargo samples that were obtained from the Grain Inspection Division of the Canadian Grain Commission (CGC). An evaluation of the computer-based methodology in terms of its ability to objectively distinguish between the three CWRS grades is also undertaken. A preliminary experiment to determine an appropriate sample size revealed that the degree of variability within the sample and the tolerable measurement error set by the investigator are important factors to consider. Using the kernel contour length feature as an example and assuming a 95% certainty that the error of estimation does not exceed 0.10mm, the results indicated that the required sample sizes are 333, 362 and 416 kernels for the 1CWRS, 2CWRS and 3CWRS grade, respectively. The average coefficient of variation (C.V.) for 16 morphological features in the carlot samples progressively increased as grade dropped, from 3.33% for the No.1 to 3.67% for the No.2 and to 4.13% for the No.3 grade. The No.3 grade was also the least uniform among the cargo grades, reflecting the fact that the highest levels of weather-related degrading factors such as bleached, immature, frosted and sprouted kernels are allowed in this grade. Each cargo grade was also found to be considerably more uniform compared to its corresponding carlot grade, with average C.V.'s decreasing by 58%, 63% and 64% for the No.1, No.2 and No.3 grade, respectively. Similar results were obtained from the thousand kernel weight determinations, providing objective evidence that Canada's grain grading and bulk handling system is very effective in enhancing uniformity within the top grades of CWRS wheat as it is moved into export position. The observed uniformity differences between the cargo and carlot grades are also indicative of the more stringent grade specifications under the export standard as compared to the primary standard.