葉菜類蔬菜硝酸鹽含量光譜影像檢測系統及栽培策略之建立

Abstract

葉菜類蔬菜為國人飲食中不可缺少的一部份，但其殘留之硝酸鹽卻不適合人體過量食用。目前量測硝酸鹽含量的方法多為精準度高但需破壞樣本之化學檢測，因此本研究希望應用光譜影像技術，利用其即時、非破壞性且具有空間資訊之優點，用以檢測蔬菜之硝酸鹽含量。另外若能於栽培期間藉由調控栽培環境使作物之硝酸鹽含量降低至標準值內，可達到消費者與栽種者雙贏的局面。 本研究分為兩大部分，一為開發出葉菜類蔬菜硝酸鹽含量即時且非破壞性之光譜影像檢測系統，二為建立以硝酸鹽含量為考量之皺葉萵苣的栽培策略模式。在非破壞性之檢測系統方面，以高光譜影像系統檢測葉片中硝酸鹽含量，並搭配MLR迴歸分析，研究結果顯示在700、450、720、950、940、470、560、550與540 nm 波長下，其rc為0.81，而SEC為1384 ppm。依據高光譜分析結果，挑選460、550、700與945 nm作為多光譜影像系統之特徵波長，分析結果顯示其rc可達0.83。相形之下，多光譜影像系統只需數個波長，且與高光譜影像系統相比需要較少影像擷取時間，因此其更適用於線上型光譜影像檢測系統。 在栽培策略模式方面，本研究針對收穫前7天之皺葉萵苣先行量測其硝酸鹽含量範圍，再以反應曲面法進行試驗之設計與分析，試驗因子包括養液氮含量、光強度與光週期。分析結果顯示，在硝酸鹽含量為2500~3000 ppm的植株，最適處理之養液氮含量濃度為101.16 ppm，光週期為日17.79 hr/夜6.21 hr，光強度為111.17 μmol/m2s；而硝酸鹽含量為3000~3500 ppm的植株，最適處理之氮含量、光週期及光強度分別為45.19 ppm、日12.71 hr/夜11.29 hr及180 μmol/m2s。由結果來看，以反應曲面法用於建立低硝酸鹽蔬菜栽培之回饋為可行的方法，可依循此模式建立更多硝酸鹽含量範圍之最適栽培模式，並針對各模式進行驗證以建立更加完整之皺葉萵苣低硝酸鹽含量回饋系統。

Vegetables are an indispensable part of daily foods in Taiwan, but excess nitrate which remnant in the vegetable is harmful to human health. The most commonly used methods for measurement nitrate measurements were chemical detections which were accurate but destructive. Therefore, this study aimed to use spectral imaging technology, which has advantages of being real-time, non-destructive and spatial informative for detecting the nitrate content of vegetables. Besides, to benefit both consumers and farmers, this study also tried to reduce the nitrate content of the crops within the required level by controlling the environment conditions during the cultivation period. This study was divided into two parts, the first one was the system development of the nitrate content of leafy vegetables immediate and non-destructive spectral imaging system; and the second part was to establish the lettuce cultivation strategies for lower nitrate contents. First, the nitrate content of lettuce leaves was detected by hyperspectral imaging system and analyzed by multiple linear regression (MLR). The results showed that while the wavelength 700, 450, 720, 950, 940, 470, 560 ,550 and 540 nm were chosen, that rc= 0.81 and SEC was 1384 ppm. According to the results, the wavelength 460, 550, 700 and 945 nm were then selected as the characteristic wavelengths of the multi-spectral imaging system, whose results gave that rc = 0.83. Compared with the hyperspectral imaging system, multi-spectral imaging system only needs a few wavelengths and requires less time of capturing images. Therefore, it was more suitable for online spectral imaging system. In this study, the response surface method (RSM) was applied to establish the feedback cultivation strategy during the cultivation which was seven days prior to harvest for the lower nitrate contents of lettuce vegetables. The analysis factors included nitrogen concentration of nutrient solution, photo period and light intensity. The results showed that the optimum cultivation strategy for the lettuces which the nitrate content between 2500 to 3000 ppm was nitrogen concentration of 101.16 ppm, the photo period of day 17.79 hr/night 6.21 hr, light intensity of 111.17 μmol/m2s. The optimum cultivation strategy for the lettuces which the nitrate content between 3000 to 3500 ppm was nitrogen concentration of 45.19 ppm, the photo period of day 12.71 hr/night 11.29 hr, light intensity of 180 μmol/m2s. As a result, RSM could be used to establish the feedback strategy during the cultivation for better control of low nitrate content in vegetables, and more optimum cultivation strategies of variable nitrate content ranges could be established by the methods developed in this study.