Combination of Laser Doppler Vibrometry (LDV) and Vis/NIRs to Predict the Firmness of Crown Pears

With the improvement of people's living standards, consumers' requirements for fruit quality are constantly rising. Firmness, which is strongly associated with the maturity of fruits, is one of the most concerned fruit quality parameters. This study aimed to utilize Laser Doppler Vibrometry (LDV) and Visible and Near-Infrared Spectroscopy (Vis/NIRs) for the non-destructive testing of the firmness of crown pears. Multiple pear firmness prediction models based on LDV and Vis/NIRs data were developed employing partial least squares (PLS), support vector regression (SVR), random forest (RF), and one-dimensional convolutional neural network (1D-CNN), and the prediction effects of these models were compared and analyzed. The experimental results demonstrate that the prediction models combining LDV and Vis/NIRs spectral data had significant advantages in accuracy compared with the prediction models using single LDV or Vis/NIRs data, and the 1D-CNN prediction model built with the fusion of the two types of data as inputs have the best prediction results (R2 P = 0.927, RMSEP = 0.678N/mm, RPDP = 3.378). This study demonstrates the brilliant performance and massive potential of the fusion of LDV and Vis/NIRs data in the application of fruit firmness prediction.