Non-Destructive Detection of Fruit Quality Parameters Using Hyperspectral Imaging, Multiple Regression Analysis and Artificial Intelligence

Currently, destructive methods are often used to measure the quality parameters of agricultural products. These complex, time consuming and costly. Recently, studying find a solution disadvantages has become major challenge for researchers. Non-destructive can be useful rapid detection In this study, hyperspectral imaging was evaluate non-destructive Red Delicious (Red Delicious) Golden (Golden apples, including pH, soluble solids content (SSC), titratable acid (TA) total phenol (TP). order predict characteristics partial least squares (PLS) method with different pre-processing used. The developed models were evaluated using root mean square RMSECV validation error, correlation coefficient (Rcv) standard deviation ratio (SDR). results showed that in Delicious, TA, SSC TP best forecasting SNV, MSC normalized regression values 0.9919, 0.9939, 0.9909 0.9899, respectively. Delicious), TP, first derivative, (smoothing second derivative), normalize (and SNV normalize) preprocessors selected as prediction models, 0.9989, 0.9999 related an artificial neural network also imaging, state feed-forward structure LM training algorithm R = 0.93, Performance 0.005 RMSE 0.03 325 inputs, 5 outputs 2 hidden layers. predictive capabilities qualitative studied study high accuracy.