數位影像處理在蔬果大小選別之應用

A Study on the Size Sorting of Vegetables and Fruits Using Digital Image Processing

謝青霖(C. L. Hsieh) ； 馮丁樹(D. S. Fon) ； 陳世銘(Su-Ming Chen)

《農業機械學刊》 1卷1期 (1992/01) Pp. 28-41

https://doi.org/10.30062/JAM.199201.0004

摘要

本研究的目的在於應用數位影像處理的理論及相關技術對蔬果大小，進行二維選別。
實驗中所採用之硬體影像處理器為The Computer Research Company所製造的Oculus 150二値化數位元影像處理介面卡，其配合之軟體則為Borland公司的Turbo C 2.0版作為C程式語言之編譯器。
實驗所得，在良好的採光之下，本研究所探討之三種理論均有良好之精確度；而在採光不良時，線段塡佈理論則可獲得較高的信賴度。
農產品品質指標之建立取決於樣品本身之物理性質、影像攝取的方法以及選別所依據之理論。實驗中發現，以線性塡佈理論建立蔬果品質指標時，Oculus 150數位影像處理器之檻値應設定在較小的範圍內以取得較精確之二維選別依據。

關鍵字

影像處理 ； 蔬果選別

並列摘要

The purpose of this study is to apply the image processing technique and theorem for the two-dimensional sorting of fruits and vegetables. The Oculus 150 frame grabber, manufactured by the COmputer Research COmpany, was used in this study. The software Turbo C 2.0 from Borland company was mainly employed to compile the C programs involved.
Among the three theorems disscussed in this study, we found that all three will obtain a better accuracy as long as the object is properly illuminated. If the illumination on the object is poor, only the line-adjacent-fill method has a higher reliability.
Several factors such as physical properties of samples, image grabbing methods, the theorem employed and etc. could have an influence on the grading index for the farm product. It was shown that, as the line-adjacent-fill method was employed to determine the grading index for fruits and vegetables, the threshold of the Oculus 150 frame grabber should be adjusted within a low range for a two-dimensional sorting with a high accuracy.

並列關鍵字

Image Processing ； Sorting of Vegetables