應用機器視覺進行番茄顏色分級之研究

Color Grading of Tomatoes with Machine Vision

楊清富(Ching-Fuh Yang) ； 李芳繁(Fang-Fan Lee)

《農業機械學刊》 3卷1期 (1994/03) Pp. 15-30

https://doi.org/10.30062/JAM.199403.0002

摘要

應用彩色影像處理系統配合分級機構及平行出料處理，發展出一套批次處理小番茄之顏色分級系統。本研究使用三種顏色系統建立了三個適合分級小番茄的顏色模式，能將小番茄分成紅色、紅綠色與綠色三個等級。分級時可藉學習的程序建立分級標準的左値和右値，當番茄顏色由機器視覺系統計算出來的値小於左値即判定為紅色等級，大於右値就判定為綠色等級，而介於兩者之間則判定為紅綠色等級。
分級結果係依照小番茄在影像平面Y方向的位置與顏色判定的等級，指向等級向量表而形成分級位元組。將分級位元組輸入分級電路在解碼之後控制分級機構產生動作，使其能夠同時處理四個番茄的分級。
三個顏色模式分級之結果與人工分級結果相符的程度分別為95.3%、93.9%及88.9%。分級作業速度最快為每分鐘380個番茄。

關鍵字

機器視覺 ； 顏色分級 ； 平行出料器 ； 等級向量表

並列摘要

A color machine vision system equipped with a grading mechanism and a parallel discharger was developed to grade cherry tomatoes by color. Three color models were established from RGB, HSI and YIQ color systems. Three color grades, namely, red, red-green, and green were used in this study. The right value and left value of the grading standards were obtained from the learning procedure. The value of the color of tomatoes was calculated by the machine vision system. It was compared with the grading standards to classify the color grade of tomatoes. If the computed color value of the tomato was smaller than the left value, the tomato was classified as red; if the color value was greater than the right value, the tomato would be graded as green; and if the color value was between the left value and the right value, then the tomato would be classified as red-green.
A grading byte was formed according to the image position of the tomatoes on the Y axis of the frame and the classified color grade of the tomatoes. The grading byte was outputted to the grading circuit to control the grading mechanism after decoding. And four tomatoes could be graded simultaneously by the system.
The grading results of the machine vision system agree with that of the manual method at 95.3%, 93.9%, and 88.9% for the H, the Q, and the G models, respectively. The maximum grading capacity of this system is 380 tomatoes per minute.

並列關鍵字

Machine vision ； Color grading ； Parallel discharger ； Grading vector table