Development of on-line detection system for simultaneous assessment of edible quality and internal defect in apple by NIR transmittance spectroscopy

Abstract. In order to simultaneous nondestructive on-line inspect edible quality and internal defect of apple, this work presents the development of an on-line detection prototype system using near infrared transmittance technology as a novel approach for on-line detect quality attributes without sample destructiveness. The on-line detection system was designed and developed to improve spectra signal quality, lower heat damage, reduce mechanical damage. Special detection software was developed for real-time inspection based on multithread programming technology. This prototype on-line system needed 80 ms capture transmittance spectra of an apple. Linear discriminant analysis (LDA) model was developed to identify internal defects samples. The results obtained from LDA models, in validation, gave a positive predictive value of classification about 90.14%. Moreover, established model strategy is first selection characteristic spectral ranges by synergy interval partial least squares (siPLS), then on the extraction of feature wavelength by stepwise multiple linear regression (SMLR). Calibration models were performed applying siPLS-SMLR algorithm to predict the soluble solid content (SSC) in apple. Results of the siPLS-SMLR model using 9 selected feature wavelengths were correlation coefficient (R) of 0.8764 and root mean square error of prediction (RMSEP) of 0.8649 Brix, respectively.. The results showed that the e on-line detection prototype based on NIR transmittance technique was feasible to simultaneous inspect the edible quality and internal defect of apple. The present research provides the foundation for the future development of an automatic system based on transmittance spectroscopy which is extremely important from the economic point of view.