蔬菜品质和新鲜度的高低不仅影响食用时的口感, 而且营养程度也不一样。 作为蔬菜品质和新鲜度重要参考指标之一的叶绿素和含水量的检测, 已经越来越受到国内外学者的重视。 相比于传统的肉眼目视判断的检验方法, 可见-近红外光谱分析具有快速高效、 无损、 非接触等独特的优势, 更加适合蔬菜的实时检测。 目前相关研究主要集中在生长中植被叶绿素和含水量的反演, 对市场上成品蔬菜的研究较少, 或者研究对象单一, 缺乏市场普适性。 此外, 光谱数据的获取需要专业的光谱仪采集, 费时费力, 各种生理生化指标的研究离实用化还有很长的距离。 为了与实际相结合, 基于智能手机光谱系统(SCSS)建立了快速、 准确、 普适性强的反演蔬菜叶绿素和含水量的模型, 并通过地面光谱仪SVC数据验证了该系统的可靠性。 选取市场典型的五种蔬菜(菠菜、 小油菜、 油麦菜、 生菜和娃娃菜)作为实验样本, 分别进行常温保存和冷藏保存来模拟现实中菜市场和超市的蔬菜储存环境。 每隔24 h进行一次数据采集。 对获取的原始光谱数据进行波段选择和小波变换去噪的预处理。 构建蔬菜叶绿素反演指数(VCRI_{( m, n)})和蔬菜含水量反演指数(VWRI_{( i, j)}), 分别提取该两个指数与叶绿素和含水量实测值的相关系数R作为权重系数, 最终建立了叶绿素和含水量的加权平均反演模型。 实验结果表明, SVC仪器和SCSS两者数据针对蔬菜叶绿素和含水量的敏感波段基本一致, 叶绿素反演的敏感波段在730980 nm之间, 反演精度 R²分别为0.863和0.808 1, 标准差为8.679 5和8.892 5; 含水量反演的敏感波段在水汽吸收波段9501 000 nm之间, 反演精度 R²分别为0.742 9和0.712 9, 标准差为8.789 9%和8.861 4%。 SVC实验数据跟SCSS实验数据结果十分接近, 验证了新型智能手机光谱系统实时监测蔬菜叶绿素和含水量的有效性。 智能手机光谱系统具有体积小、 价格便宜的优势, 结合网络云端服务和实时数据反馈的特点, 能够实现蔬菜品质和新鲜度指标的智能检测, 让光谱分析真正应用于人们日常生活中。

The quality and freshness of vegetables not only affect the taste, but the nutrient content. The research on detection of chlorophyll and water content that are important reference indexes of vegetables quality and freshness has become more attention to the researchers both at home and abroad. With the quickness, high efficiency, non-destruction and non-contact features, the novel visible/near-infrared spectral analysis technology is more suitable for real-time detection of vegetables, comparing wtih traditional estimating methods by naked eyes. The relevant research is primarily focus on retrieval of growing vegetation chlorophyll and water content at present. There is little research aiming at ripe vegetables in market, or lacking of universality because of single species. Moreover, collecting of spectral data requires professional Field Spectrometer, wasting time and energy. There is a distance between research of physiological and biochemical index and practical application. In order to combining the research with the real life, this paper builds quickly, precise, universal models that can retrieve chlorophyll and water content in vegetables, based on Smart Cellphone Spectral System(SCSS). Simultaneously, SVC are used to validate the reliability of SCSS. Five kinds of common vegetables (spinach, rape, romaine, lettuce and baby cabbage) are selected as samples in experiment, and the ways of cold storage and normal temperature preservation are used to simulate the market and supermarket environment. Datas are collected per 24 hours. Then Band-Selecting and Wavelet-Transform preprocessing are adopted to improve the quality of spectral data. This paper constructs Vegetable Chlorophyll Retrieval Index (VCRI) and Vegetable Water Retrieval Index (VWRI), and extracts the correlation coefficients between the two indexes and measured values of chlorophyll and water content as weight coefficients. Finally, the chlorophyll and water content retrieval models are built. The result shows, SVC and SCSS have the same sensitive bands to chlorophyll and water content. The sensitive wavelength for chlorophyll retrieval is from 730 to 980 nm. The precision R² are 0.863 and 0.8081, and standard deviation are 8.679 5 and 8.892 5 respectively. The sensitive wavelength for water content retrieval is from 950 to 1 000 nm. The precision R² are 0.742 9 and 0.712 9, and standard deviation are 8.789 9% and 8.861 4% respectively. The result of SVC and SCSS is similar enough to prove the validation of new-style Smart Cellphone Spectral System. Furthermore, SCSS has the advantage of small size and low price. It can smartly detect the quality and freshness index of vegetables, with the features of internet cloud services and data feedback in real-time. This makes the spectral analysis technology applying to the people’s daily life.