行政院農業委員會台南區農業改良場   研究彙報第50號

 

文心蘭切花影像分級技術之研究

楊清富、鄭榮瑞、鍾瑞永

摘  要

楊清富、鄭榮瑞、鍾瑞永•2007•文心蘭切花影像分級技術之研究。台南區農業改良場研究彙報50:31 - 38。 為提升文心蘭切花採收後處理效率與品質,自動化處理技術之開發為不可或缺的一環。現行文心蘭切花全憑人工目測進行分級,因此分級效率不易提升且等級判定之準確度亦未臻客觀。本研究之目的旨在建立一套適合文心蘭切花分級之影像技術俾建構自動化作業處理系統,以降低人工成本並提高作業效率。在本研究利用彩色攝影機及影像卡擷取整支切花影像,在紅色成份影像中以兩個影像門檻值有效的將文心蘭花朵與莖部影像分離,並進行花朵區塊辨識及花朵像素計算,再以切割法搜尋切花第一分叉點而計算出切花花部及花莖長度。切花等級利用花朵總像素取代分叉數、花部長、莖部長等條件配合現行分級標準進行判定,分級結果與人工分級相符程度為82%,判別一支切花平均耗時70ms。 關鍵詞:文心蘭、影像處理 接受日期:2008年1月7日


Research of Grading Oncidium Cut Flower by Image Processing

Yang C.F. , J.J. Cheng and J.Y. Chung


Summary

In order to elevate the processing efficiency and quality of post harvest technology for Oncidium cut flower, the development of automotive processing technology is quite vital and may not be short of.? Currently, we mainly adopt artificial visual inspection to grade Oncidium cut flowers.? Therefore, its efficiency is not easy to be lifted and the accuracy of judgment is not so objective.
The purpose of this research is to build automotive standard operation process by Oncidium cut flower grading image processing to reduce labor cost and enhance operation efficiency.
This research uses color camera and image acquisition card to grab the whole cut flower image.? In Red Frame image, we separate efficiently Oncidium flower & its stem with two thresholds.? Further, we do flower blob identification and flower Pixel calculation.? And then, we use slicing method to search for the first crotched point to calculate the cut flower substance and the length of the stem.? The cutting flower classification makes use of the flower total pixels instead of crotch ?number, flower substance and stem length…and the other conditions to meet the current grading standard and make judgment.? The result of grading, to compare with the labor grading, may lead to 82%.? It costs 70 ms. for one cut flower on average.
Keywords:Oncidium、Image Processing
Accepted for publication:7 January, 2008


    回研究彙報目錄