DETECTION OF HOLLOW PEARS BY TREE BASED MODELLING ON NON-DESTRUCTIVE ACOUSTIC RESPONSE SPECTRA

Abstract

After CA storage, some pears show internal breakdown and brown discoloration. This can be caused by a combination of picking time and storage conditions. There is no visual evidence on the outside of the fruit that there is an internal defect. When the quality of horticultural products is a major issue, it becomes a problem for marketing strategies when the consumer finds hollow pears. Acoustic response spectra have already proven to be a valuable tool in the determination of some internal characteristics of fruits, like the stiffness of tomatoes. The acoustic response spectra show some abnormalities for pears with internal cavities. Classical numerical values (second moment, damping, frequency etc.) based on these spectra, were used to identify non-marketable fruits. An additional Fourier series algorithm of the spectrum and the use of tree based modelling on the coefficients yielded satisfactory results. This made it possible to identify the hollow pears quite adequately. Since the protocol is based on non-destructive methods, it is possible to incorporate it in a sorting line. In this way a substantial amount of affected pears can be removed, resulting in better overall quality.