Predicting the quality attributes related to geographical growing regions in red-fleshed kiwifruit by data fusion of electronic nose and computer vision systems

Abstract

The ability of a data fusion system composed of a computer vision system (CVS) and an electronic nose (e-nose) was evaluated to predict key physiochemical attributes and distinguish red-fleshed kiwifruit produced in three distinct regions in northern Iran. Color and morphological features from whole and middle-cut kiwifruits, along with the maximum responses of the 13 metal oxide semiconductor (MOS) sensors of an e-nose system, were used as inputs to the data fusion system. Principal component analysis (PCA) revealed that the first two principal components (PCs) extracted from the e-nose features could effectively differentiate kiwifruit samples from different regions. The PCA-SVM algorithm achieved a 93.33% classification rate for kiwifruits from three regions based on data from individual e-nose and CVS. Data fusion increased the classification rate of the SVM model to 100% and improved the performance of Support Vector Regression (SVR) for predicting physiochemical indices of kiwifruits compared to individual systems. The data fusion-based PCA-SVR models achieved validation R² values ranging from 90.17% for the Brix-Acid Ratio (BAR) to 98.57% for pH prediction. These results demonstrate the high potential of fusing artificial visual and olfactory systems for quality monitoring and identifying the geographical growing regions of kiwifruits.

Keywords: Image processing; Machine learning; Origin; Physicochemical attributes; Volatile Organic compounds.

Fig. 1

Climatic data of the three studied regions: a Temperature, b Rainfall [34]

Fig. 2

Actual views of the used computer vision (a) and e-nose (b) systems

Fig. 3

Flowchart of image processing and feature extraction from images of whole kiwifruits

Fig. 4

Flowchart of image processing and feature extraction from images of middle-cut section of kiwifruits

Fig. 5

Gallery of whole fruit images at different steps of image processing; a RGB image of whole kiwifruit, b gray-scale image, c binary image of whole kiwifruit, d The whole kiwifruit RGB image resulted by applying logical AND between RGB image and binary image, e converted to HSV color space, and f) converted to L*a*b* color space

Fig. 6

Gallery of middle-cut images at different steps of image processing; a RGB image of kiwifruit middle cut, b H component, c binary image of fruit middle-cut, d ExG image, e binary image of outer region, f RGB image of outer region, g binary image of core region, h RGB image of core region, i binary image of locule region, j RGB image of locule region, and k) colored image of segmented image

Fig. 7

Radar plot of the MSR averages of e-nose sensors for red-fleshed kiwifruits of different growing region

Fig. 8

PCA score (a) and loading (b) plots for growing region discrimination of kiwifruit using e-nose data

Fig. 9

PCA score (a, c) and loading (b, d) plots for growing region discrimination of kiwifruit using image-extracted features from whole fruits (a, b) and middle-cult fruits (c, d)

Fig. 10

Results of data fusion-based PCA-SVR models for prediction of kiwifruit firmness (a), SSC (b), TA (c), and BAR (d)

Fig. 11

Results of data fusion-based PCA-SVR models for prediction of kiwifruit pH (a), vitamin C (b), TP (c), TAC (d), DPPH (e), and FRAP (f)

Fig. 12

PC importance plots of PCA-SVR models for prediction of kiwifruit firmness (a), SSC (b), TA (c), BAR (d), pH (e), vitamin C (f), TP (g), TAC (h), DPPH (i), and FRAP (j)