基于高光谱成像技术的番茄内部品质检测方法研究
打开文本图片集
中图分类号：TP391.4；S436.421 文献标识码：A 文章编号：2095-5553（2026）04-0205-09
Abstract：Tobreak throughthe limitationsof traditional destructive detectionmethodsandachievearapid， non-destructiveand precisenew method for detecting the internal qualityof tomatoes，providing moreeficient technical means for thequalityasessmentof agricultural products.This paper is basedon hyper spectral imaging technology and machine learning methods to realizenon-destructive detection of the soluble solidcontent（SSC）of tomatoes.Hyper spectral data of tomato samples were collected，and four preprocessing methods，including multiplescatter correction （MSC），standard normal variate transformation（SNV），Z—Score standardization，andMin—Max normalization，were adopted.Thecorresponding PLSR modelswere established，and the PLSR model basedon Min—Max normalization preprocessing achieved the best results，with RPD increasing to 3.282 4 and RMSE being 0.425 3. SPA，CARS，and UVEalgorithms wereused toextractfeature variables from theMin—Max normalized spectral data，andthefinal number of extracted feature variables was 27，33，and 88，respectively.PLSR，BPNN，and SVMR prediction models for tomato SSC were established based onthese three feature extraction methods.Theresults show that the SPA—SVMR model hasthe best prediction efecton thesoluble solidcontent in theinternal qualityof tomatoes，withthe corresponding calibration set coefficient of determination Rc2 and prediction set coeficient of determination Rp2 reaching 0.6454and 0.605O，respectivelyThe root mean square erorof thecalibration set RMSECand the root mean square errorof the prediction set RMSEP were O.334 2 and O.4Ol 9，respectively，and the relative prediction deviation RPD of the prediction set was 3.9588.The results indicate thatthecombination of hyper spectral technologyand machine learning can achieve rapid and non-destructive prediction and analysis of the internal quality of tomatoes.
Keywords：hyperspectral imaging technology；machine learning；tomato；non-destructive detection；soluble solids content
0 引言
可溶性固形物含量（SSC）是番茄品质评价的核心指标之一，具体是指番茄汁液中诸如可溶性糖、有机酸、番茄红素等溶质所占的百分比[]。（剩余14114字）
