水果品质检测分级是水果产后商品化处理的重要关节，基于可见/近红外光谱的水果内部品质无损检测是近10多年来的研究热点，但由于在实际应用中面临两大因素的影响：第一是受样品相关因素的扰动，第二是受仪器与环境等因素的影响，长期以来如何建立水果有效光谱信息的获取方法并提高预测模型的稳健性一直是本领域的研究难点问题。为此，本项目拟开展水果有效光谱信息在线实时获取方法及其糖度无损检测模型的创新研究：（1）研究光谱检测系统的优化；（2）研究样品光谱和参比光谱最佳采集组合模式；（3）研究样品相关因素对预测模型的影响；（4）研究特征波长选择方法及其模型对新样本的预测能力。本项目将建立一种水果有效光谱信息在线实时获取方法及其糖度预测模型，对推进可见/近红外光谱实时无损检测技术在我国水果产后商品化处理中的应用具有重要的意义。

Fruit quality inspecting and grading is an important issue in fruit postharvest commercialization processing. Research about nondestructive inspection of fruit internal quality based on visible/near infrared (Vis/NIR) spectroscopy is popular worldwide for more than 10 years. However, there are two obstacles in practical application of this technique. The first is the disturbance caused by sample-related factors. The second is the influences of instrumental and environmental factors. It has been a difficulty of how to obtain effective spectra of fruits and establish robust prediction models. Therefore, this project is aimed to do innovative study on methods for on-line real-time acquisition of effective spectra and models for nondestructive sugar content detection of fruits: (1) study on optimization of spectral inspection system; (2) study on optimization of combination mode for collecting sample spectra and reference spectra; (3) study on the effects of different sample-related factors on prediction models; (4) study on methods for characteristic wavelength selection and model prediction performance for new samples. This project will establish a method to obtain effective spectra of fruits and to set up robust prediction models , which has great significances for promoting the applications of the real-time nondestructive inspection technique based on Vis/NIR spectroscopy for fruit postharvest commercialization processing in our country.