Crispness is a vitally significant factor to evaluate apple texture quality. This study explores quantitative prediction and kinetic models to determine the optimal preservation time of apple crispness during shelf life based on spectroscopy. The nonlinear iterative partial least squares algorithm was used to establish a quantitative prediction model based on the wavelength of 450–1000 nm. Kinetic models were developed to determine the preservation time of apple crispness at room and refrigeration temperatures. The results indicate that the determination coefficients of calibration and prediction sets were 0.8939 and 0.9206 respectively, and the root mean square errors of calibration and prediction sets were 0.1254 and 0.1669 kg, respectively. The determination coefficients of the kinetic models at room and refrigeration temperatures were 0.965 and 0.87, respectively. Consequently, the preservation time of the optimal freshness of Fuji apple crispness was five weeks at room temperature and eight weeks at refrigeration temperature. Practical applications This study provides a method for nondestructively and accurately detecting apple crispness via spectroscopy. This study also provides optimal freshness taste of apple crispness and how long apple crispness could retain under the conditions of room and refrigeration temperature during shelf life for customers and academics. The corporate could extend the preservation time according to this study to improve the profile.

Crispness is regarded as a significant quality index for apples. Currently, destructive sensory evaluation is the accepted method used to detect apple crispness, making it essential to develop a method that can detect apple crispness in a nondestructive manner. In this study, spectroscopy was proposed as the nondestructive technique for detecting apples' crispness, ultimately obtaining a spectral reflectance curve between 450 nm and 1,000 nm. In order to simplify and improve modeling efficiency, successive projections algorithm (SPA) and x‐loading weights (x‐LW) methods were used to select the most effective wavelengths. Partial least squares (PLS) algorithm, radial basis neural networks (RBNN), and multilayer perceptron neural networks (MLPNN) methods were used to establish the models and to predict the crispness of “Fuji” and “Qinguan” apple varieties. Based on the full wavelength (FW), the prediction accuracy of the PLS model for “Fuji” and “Qinguan” apple varieties was 92.05% and 95.87%, respectively. The effective wavelengths selected via SPA for the “Fuji” apple variety were 450.41 nm, 476.80 nm, 677.75 nm, and 750.72 nm, and the effective wavelengths selected via x‐LW for the “Qinguan” apple variety were 542.51 nm, 544.79 nm, 676.96 nm, and 718.29 nm. The prediction accuracy of the PLS model based on effective wavelengths for “Fuji” and “Qinguan” apple varieties reached 91.31% and 96.41%, respectively. Compared with the RBNN model, the MLPNN model achieved better prediction results for both “Fuji” and “Qinguan” apples, with the prediction accuracy reaching 97.8% and 99.9%, respectively. Based on the above findings, effective wavelength selection and MLPNN modeling were able to detect apple crispness with the highest accuracy. Overall, it can be concluded that the less effective wavelengths are conducive to developing an instrument for crispness detection.

No abstract

Texture is an important attribute affecting apple quality and consumer preferences.In general, apple texture is evaluated by such parameters as crispness and hardness.To explore the feasibility of nondestructive detection of apple crispness based on spectroscopy and machine vision, a calibration model based on optical fiber spectroscopy (in the range of 500-1,000 nm) was developed using the partial least squares. A compensation model based on appearance images was established to improve the prediction performance, and a quantitative prediction model based on the fusion of spectra and appearance images was subsequently obtained. The results showed that the determination coefficient of prediction (R 2 p ) and root mean square error (RMSE) of prediction (RMSEP) of the calibration model were 0.8921 and 0.1226 N, respectively. The determination coefficients (R 2 ) of the compensation model for positive derivation and negative derivation were 0.9310 and 0.9051, respectively, with a RMSE of 0.0515 N and 0.0420 N, respectively. The mean value and standard deviation that occurred in relative errors both decreased from 8.52 to 6.44% and 6.22 to 5.30% before and after positive derivation compensation and decreased from 8.29 to 5.55% and 5.62 to 4.49% before and after negative derivation compensation using the appearance images. This result indicated that the prediction of apple crispness was more accurate and reliable after the fusion of spectra and appearance images.Therefore, it is feasible to nondestructively detect apple crispness based on spectroscopy and machine vision. Practical ApplicationsThis study provides a method for nondestructively and accurately detecting apple crispness via spectroscopy and machine vision. According to the crispness of apples, enterprises can not only sell apples to different consumer groups, but also distribute apples in different levels to improve their benefits.