Characteristic information analysis of Raman spectrum of cucumber chlorophyll content and hardness and detection model construction

The study of the rapid detection method of cucumber chlorophyll content and hardness can lay the foundation for the rapid detection of cucumber quality. A Raman Spectroscopy coupled with Competitive Adaptive Reweighted Sampling (CARS) and Support Vector Machines (SVM) method was reported to detect chlorophyll content and hardness of cucumber under different storage time. Mapping images of chlorophyll content and hardness were constructed based on Raman wavenumbers within the range of 1500–1550 cm^− 1and1300-1350 cm^− 1, respectively. Distribution and changes of the chlorophyll content and hardness were demonstrated by mapping images. 35 and 33 characteristic wavenumbers for chlorophyll content and hardness of cucumber were extracted by CARS, respectively. The full spectral wavenumbers, characteristic wavenumbers of mapping image and characteristic wavenumbers extracted by CARS were compared, and 4 kind of kernel functions were analyzed. For the chlorophyll and hardness, support vector machine (SVM) model constructed based on the CARS characteristic wavenumbers and RBF kernel function had achieved good prediction result, The R2 of training set and the testing set was both over 0.9. The results showed that Raman spectroscopy technology combined with the CARS-SVM algorithm can achieve a rapid detection of chlorophyll content and hardness in cucumber.