- 期刊
- OpenAccess
糙米外觀品質影像特徵之研究
Image Characteristics of Brown Rice Appearance Quality
摘要
本研究應用機械視覺建立了判定糙米外觀品質之量化參數，包括打光及米粒取像環境的設定方式、米粒影像品質特徵之量化、鑑別值的計算、和糙米外觀品質判定參數之統計選取模式。在以台農67號糙米爲例的研究，計算出糙米投影面積、米粒上下端部寬度值、白堊值所佔米粒面積比例、米粒長短軸比、米粒表面紅色光度値和米粒表面綠色光度值等六個參數有最佳效果，可將糙米依其外觀分爲畸型粒、著色粒、白死米、褐米、發芽粒、青未熟粒、青死米、正常粒、基部未熟粒、其他未熟粒、白堊質粒與破碎粒等十二種米粒。這些參數可以作爲育種、作物生長研究、病蟲害防治等研究時，一個客觀，可量化之稻米檢測的依據，以取代主觀的人工檢測。本研究所建立的糙米外觀品質檢測參數，如進一步配合品級判別程式的發展，將可作爲自動檢測糙米之CNS等級之用。
並列摘要
In this study, machine vision is applied to create quantitative parameters for inspecting outward appearance quality of brown rice. These include lighting and environment setting to take rice kernel image, quantifying quality characters from a kernel image, threshold estimation, and applying statistic almethod to find parameters which are proper for quality inspection of brow nrice. From the example study of ”Tainung 67” brown rice, computation results show parameters of projection area, mean width of top and bottom, chalkyarea ratio, ratio of long and short axles, average gray level of red color component, and average gray level of green color component six parameters from a kernel image have the best effects. They can be applied to classify brown rice kernel according to their appearance into abnormal, discolored, white dead, rusty, sprouted, green immature, green dead, sound, base immature, other immature, chalky and broken, totally twelve categories. These parameters can also help researchers in crop growth studies, diseases and insect's prevention et al. This is an objective rice quality inspection with measured value instead of subjective quality inspection by person. All these quality inspection parameter sestablished in this study, if combined with a self grading program, can apply to grade brown rice automatically according to CNS standard.