Non-destructive Detection of Blackheart Potatoes Based on Energy Spectrum of VIS/NIR Transmittance

Abstract

Abstract Blackheart disease can cause internal defects in potatoes, which seriously damages the quality and yield of processed potato products and causes considerable economic losses. To achieve online grading of potato, it is important to develop an accurate and reliable method to identify and separate blackheart potatoes non-destructively. A new method of transmitted energy spectrum determination was analysed and adopted innovatively for blackheart potato detection. Three methods, peak area linear discrimination analysis (PA-LDA), peak value linear discrimination analysis (PV-LDA), and peak difference value linear discrimination analysis (PDV-LDA), were used to identify blackheart potato, respectively. All of the three determination methods (PA between 657nm to 750nm normalized by T580, T698/T657, (T698-T657)/T624) could identify blackheart potatoes accurately, and the classification accuracy were 91.69%, 92.43% and 93.69%, respectively. PDV-LDA ((T698-T657)/T624) showed the best performance, whose sensitivity, specificity, and AUC value were 94.86%, 95.19%, and 0.98, respectively. Compared with the general absorbance spectrum determination method, this method has simple operation, stable model, and can realize online detection fastly and accurately.