蔬菜葉片氮含量之近紅外光反射光譜分析

Determination of Nitrogen Content in Vegetable Leaves Using NIR Reflectance Spectra Analysis

陳加增(Chia-Tseng Chen) ； 陳世銘(Su-Ming Chen) ； 楊蕙綺(Huei-Chi Yang) ； 楊宜璋(I-Chang Yang) ； 蕭世傑(Shi-Jie Hsiao)

《農業機械學刊》 15卷4期 (2006/12) Pp. 37-49

https://doi.org/10.30062/JAM.200612.0004

摘要

本研究利用近紅外光分光光度計，進行葉片粉末反射吸收光譜量測及分析，建立蔬菜葉片氮含量之光譜檢測模式；並進行不同波段光譜檢測模式之探討，以評估未來進一步發展植物養份狀態之多光譜影像遙測技術之可行性。以分光光度計NIRS 6500(FOSS NIRSystems)量測三種不同氮肥栽培之鳳京白菜植株葉片粉末之反射吸收光譜，選擇全波段(400-2500nm)、有效特徵波段(1400-2450nm)和矽質CCD攝影機感光波段(450-1000nm)等三種光譜資料，並以平滑化、平滑化併一次微分和平滑化併二次微分等三種光譜前處理，探討修正型部份最小平方迴歸(modified partial least square regression, MPLSR)和逐步式多重線性迴歸(stepwise multiple linear regression, SMLR)兩種光譜檢測模式在上述條件下之氮含量預測能力。以光譜前處理特性而言，單純的平滑化光譜處理，並無法增進光譜值與氮含量之線性關係，平滑化光譜再經由微分處理後，因粉末顆粒大小不同所造成之光譜平移干擾已被大幅減少，故在特徵波段1400-2450nm範圍內，已有多處波段之線性相關係數絕對值可達0.9以上。在氮含量光譜檢測模式分析中，以平滑化併一次微分之四波長（2124nm、2240nm、1666nm和1632nm）SMLR模式預測能力為最佳(SEC=2.059mg/g,rc=0.991,SEV=2.131mg/g,rv=0.990)，優於採用相同光譜前處理之最佳MPLSR模式，顯示僅使用少數波長建構之光譜檢測模式，在無水份光譜吸收的干擾條件下，仍可勝於MPLSR。因此，可利用此一植株葉片氮含量之光譜檢測模式，取代傳統耗時費工的濕式化學分析。此外，本研究成果指出，若進一步考慮將光譜檢測技術應用於現場植株生長之氮肥營養管理，可以採用短波近紅外光攝影機搭配液晶可調濾波器拍攝植株高光譜影像，以二次微分併平滑化之450~1000nm光譜波段資訊，建構植株高光譜影像之MPLSR氮含量檢測模式，以供生產現場管理之用。

關鍵字

蔬菜 ； 近紅外光 ； 氮含量 ； 特徵波長

並列摘要

This study aims to develop the nitrogen prediction model of vegetable leaves using near infrared spectroscopy, and to investigate the feasibility of multi-spectral image remote sensing by spectral analyses of different selected spectral bands. Chinese mustard (Brassica rapa L. var. chinensis (Rupr.) Olsson) was cultured by three different nitrogen fertilization treatments, and the reflectance spectra of leaves in terms of powder form were measured using NIRS 6500 (FOSS NIRSystems). Two models including modified partial least square regression (MPLSR) and stepwise multiple linear regression (SMLR) were developed in full band (400-2500nm), selected band (1400-2450nm), and silicon CCD sensing band (450-1000nm) with three pre-treatments, namely, smooth, smooth and first derivative, smooth and second derivative. The results showed that derivative treatments could reduce the noises of spectral shift caused by the particle size, and the significant wavelengths with high correlation coefficient (|r|＞0.9) were appeared in the selected significant spectral band (1400-2450nm). Regarding the nitrogen prediction models, SMLR with smooth and first derivative pre-treatment and four significant wavelengths (2124, 2240, 1666, and 1632nm) gave the best results (SEC=2.059mg/g, rc=0.991, SEV=2.131mg/g, rv=0.990). The results pointed out the SMLR model with a few wavelengths as inputs would be better than MPLSR when spectral information without water absorbance interference, and the SMLR model could be used to replace the time-consuming wet chemical method, such as Kjeldahl method, to analyze the nitrogen content in vegetable leaves. The results also indicated that a hyper-spectral imaging system, constructed of silicon CCD cameras and liquid crystal tunable filters (LCTFs), using MPLSR method with the smooth and second derivative spectral information in range of 450 to 1000 nm could be used for nitrogen fertilization management of vegetable growth in the field.

並列關鍵字

Vegetable ； Near Infrared ； Nitrogen Content ； Significant Wavelengths