甘藍種苗之生長量測與模式分析：（一）自動量測系統的建立

Growth Measurement and Modeling of Cabbage Seedlings: Ⅰ.Implementation of the Automatic Measurement System

鄭圣夫(Sheng-Fu Cheng) ； 林遠德(Ta-Te Lin)

《農業機械學刊》 6卷4期 (1997/12) Pp. 69-83

https://doi.org/10.30062/JAM.199712.0007

摘要

於控制環境下進行植物生長之量測，其資料有助於我們瞭解所栽培植物之生長狀態與環境因子的互動關係，因此在本研究中我們發展一套以機器視覺系統為基礎的自動化量測系統，以利進行種苗生長過程的動態量測。量測系統以甘藍種苗為實驗對象，利用背光方式取其側面影像後進行影像處理，並以旋轉機構旋轉種苗取得各角度之影像後應用程式計算種苗之外觀性狀，計算所得之性狀值再經校正方程式換算為實際單位。主要量測的影像外觀性狀計有株高、投影面積、葉片與莖部節點數，以及葉片端點與莖部節點座標。投影面積與種苗之地上部鮮重、葉總面積及地上部乾重之間有很高之相關性，因此據以建立校正方程式，此三項種苗性狀由影像投影面積，經校正方程式換算所得之量測值與人工量測值的百分誤差分別為9.0±5.2%、9.4±6.8%與10.7±7.2%。由單張影像所計算出來的葉片端點與莖部節點座標，可經由多張各角度影像的組合運算求得其實際的三維座標，再根據點與點之間的連接關係，以電腦繪圖原理可建立一簡單的三維種苗模型，提供種苗生長過程之動態展示及模式分析之應用。

關鍵字

影像處理 ； 甘藍種苗 ； 生長量測

並列摘要

The measurement of plant growth under controlled environment is he1pful in understanding the interaction between plant growth and environmental factors. Therefore, an automatic plant growth measurement system employing machine vision was developed in this research. The system was tested and applied to cabbage seedling growth measurement. Eight seedling profile images were taken from each seedling placed on a rotary stage. Images from the back-lighted seedling were first preprocessed into binary images and then the seedling characteristics were determined with image processing algorithms and calibration procedures. The measured seedling characteristics were: seedling height, projection area, leaf and stem node numbers, coordinates of stem nodes and leaf endpoints. The seedling projection area from the images was highly correlated to the fresh weight, total leaf area and dry weight. Three calibration curves were regressed for these seedling characteristics and the calibration curves were verified with a separate experiment. The relative measurement error were 9.0±5.2%, 9.4±6.8% and 10.7±7.2%, accordingly. The developed system also equipped with the capability of reconstructing the simple 3-dimensional structural model of the seedling. This was achieved by combining the information of node coordinates from eight profile images. The simple model allows for dynamic graphic display of seedling growth and application of further modeling and analysis.

並列關鍵字

Image processing ； Cabbage seedling ； Growth measurement