nondestructive detection of kiwifruit textural characteristic based on near infrared hyperspectral imaging technology

The nondestructive detection of kiwifruit texture has become an important necessity that influences economic efficiency and consumer recognition the kiwifruit. This study investigated feasibility using hyperspectral imaging technology to identify textural characteristics kiwifruit, establish model with best performance. Firstly, a near-infrared online grading system (1000–2500 nm) was constructed acquire spectral images. Textural were measured via three different analysis tests: profile analysis, puncture test, shear test. Then, samples divided into calibration set prediction based on rank sampling at ratio 3:1. Mean centralization, standard normal variate, multiplicative scatter correction methods applied pre-process obtained spectra. Finally, partial least squares method used models characteristics. hardness1, chewiness, resilience, peel hardness, average corrected hardness achieved good Both correlation coefficients (rc) (rp) values exceeded 0.9. difference between root mean square error (RMSE) small, deviation (RPD) value 2, which predict quality. However, force had low accuracy RPD 1.680 also below 1.5 RMSE. results showed technique can be as for