Non-destructive detection of kiwifruit soluble solid content based on hyperspectral and fluorescence spectral imaging

Abstract

The soluble solid content (SSC) is one of the important parameters depicting the quality, maturity and taste of fruits. This study explored hyperspectral imaging (HSI) and fluorescence spectral imaging (FSI) techniques, as well as suitable chemometric techniques to predict the SSC in kiwifruit. 90 kiwifruit samples were divided into 70 calibration sets and 20 prediction sets. The hyperspectral images of samples in the spectral range of 387 nm~1034 nm and the fluorescence spectral images in the spectral range of 400 nm~1000 nm were collected, and their regions of interest were extracted. Six spectral pre-processing techniques were used to pre-process the two spectral data, and the best pre-processing method was selected after comparing it with the predicted results. Then, five primary and three secondary feature extraction algorithms were used to extract feature variables from the pre-processed spectral data. Subsequently, three regression prediction models, i.e., the extreme learning machines (ELM), the partial least squares regression (PLSR) and the particle swarm optimization - least square support vector machine (PSO-LSSVM), were established. The prediction results were analyzed and compared further. MASS-Boss-ELM, based on fluorescence spectral imaging technique, exhibited the best prediction performance for the kiwifruit SSC, with the ${\text{R}}_p^2$ , ${\text{R}}_c^2$ and RPD of 0.8894, 0.9429 and 2.88, respectively. MASS-Boss-PLSR based on the hyperspectral imaging technique showed a slightly lower prediction performance, with the ${\text{R}}_p^2$ , ${\text{R}}_c^2$ , and RPD of 0.8717, 0.8747, and 2.89, respectively. The outcome presents that the two spectral imaging techniques are suitable for the non-destructive prediction of fruit quality. Among them, the FSI technology illustrates better prediction, providing technical support for the non-destructive detection of intrinsic fruit quality.

Keywords: fluorescence spectral; hyperspectral; kiwifruit; non-destructive detection; ssc.

Figure 1

The overall equipment structure: (A) Gaia hyperspectral sorter; (B) Gaia fluorescence spectral detection system.

Figure 2

Spectral image of a sample: (A) raw hyperspectral image; (B) raw fluorescence spectral image; (C) ROI of the raw hyperspectral image; (D) ROI of the raw fluorescence spectral image.

Figure 3

Spectral data of kiwifruit acquired by using (A) hyperspectral imaging and (B) fluorescence imaging.

Figure 4

Distribution of the feature variables extracted by Boss.

Figure 5

Extraction process of hyperspectral feature variables by CARS: (A) The number of feature variables reserved; (B) RMSECV; (C) The change of regression coefficient of each characteristic variable.

Figure 6

Extraction process of fluorescence spectral feature variables by CARS: (A) The number of feature variables reserved; (B) RMSECV; (C) The change of regression coefficient of each characteristic variable.

Figure 7

Distribution of the feature variables extracted by CARS.

Figure 8

Distribution of the feature variables extracted by IVSO.

Figure 9

Distribution of feature variables extracted by IVISSA.

Figure 10

Spectral feature variable distribution map based on MASS.

Figure 11

The number of variables extracted by different feature extraction methods.

Figure 12

Prediction results of different optimal methods: (A) hyperspectral-MASS-Boss-ELM; (B) hyperspectral-MASS-Boss-PLSR; (C) hyperspectral-IVSO-PSO-LSSVM; (D) fluorescence-MASS-Boss-ELM; (E) fluorescence-CARS-PLSR; (F) fluorescence-IVISSA-PSO-LSSVM.