摘要
【目的】鲜莲直链淀粉含量直接影响后续的加工品质,课题组提出了基于高光谱成像技术的鲜莲直链淀粉检测方法的研究。【方法】获取120颗不同品种鲜莲的高光谱图像及直链淀粉含量,对感兴趣区域(region of interest,ROI)平均光谱采用不同的预处理方法,然后建立偏最小二乘回归(partial lease square regression,PLSR)模型检测直链淀粉。通过对比选用多元散射校正(multiplicative scatter correction,MSC)和一阶导数结合预处理后的光谱作为波长优选的输入,其次分别采用回归系数法(regression coefficient,RC)和连续投影算法(successive projections algorithm,SPA)进行特征波段提取并建立PLSR模型,确定最优的建模波段。【结果】采用SPA从471个波段中优选出了449.5 nm、464.1 nm、500.6 nm、582.7 nm、644.3 nm、756.1 nm、812.4 nm、872.3 nm和977.6 nm共9个特征波段。建立的偏最小二乘模型校正集的相关系数(R_(c))为0.835,校正集均方根误差(RMSEC)为1.802,预测集的相关系数(R_(p))为0.856,预测集均方根误差(RMSEP)为1.752,相对分析误差(RPD)为1.944。【结论】基于高光谱成像技术可以实现对鲜莲直链淀粉含量的检测,可为进一步开发基于多光谱技术的莲子直链淀粉含量便携式检测设备提供参考。
[Objective]The content of amylose in fresh lotus directly affects the further processing quality.The research group proposed the research on the detection method of fresh lotus amylose based on hyperspectral imaging technology.[Methods]The high spectral images and amylose content of 120 different fresh lotus were obtained.Different pretreatments were used to cope with the average spectra obtained from the regions of interest (ROI),and then the partial least square regression (PLSR) model was established to detect amylose.Through comparison,the spectrum after multiplicative scatter correction (MSC) and the first derivative combined with preprocessing was selected as the input of wavelength optimization.Then the regression coefficient method (RC) and successive projections algorithm (SPA) are used to extract the feature band,establish the PLSR model and determine the optimal modeling band.[Results]Nine characteristic bands of 449.5 nm,464.1 nm,500.6 nm,582.7 nm,644.3 nm,756.1 nm,812.4 nm,872.3 nm and 977.6 nm were selected from 471 bands by SPA.The correlation coefficient of calibration (R_(c)) of the established partial least squares model is 0.835,the root mean square error of calibration (RMSEC) is 1.802,the correlation coefficient of prediction (R_(p)) is 0.856,the root mean square error of prediction (RMSEP) is 1.752,and the residual predictive deviation (RPD) is 1.944.[Conclusion]The detection of amylose content in fresh lotus seeds can be realized based on hyperspectral imaging technology,which can provide a reference for the further development of portable detection equipment for amylose content in lotus seeds based on multispectral technology.
作者
高胜
黄亮
邹金平
李佳育
魏萱
Gao Sheng;Huang Liang;Zou Jinping;Li Jiayu;Wei Xuan(College of Mechanical and Electronic Engineering,Fujian Agricultrure and Forestry University,Fujian Fuzhou 350100;College of Food Science,Fujian Agricultrure and Forestry University,Fujian Fuzhou 350002;Fujian Province’s Key Laboratory of Agricultural Information Sensing Technology,Fujian Fuzhou 350100)
基金
中国博士后基金项目“光学特性解析鲜莲直链淀粉检测模型传递的方法及机理”(2019M662216)
福建农林大学校级纵向项目“学科交叉融合推进智慧农业(园艺)发展”(000/71202103B)。
关键词
莲子
直链淀粉
高光谱成像
特征波长
lotus seed
amylose
hyperspectral imaging
feature wavelengths
