Internal defect detection in fruit by using NIR spectroscopy

Retailers commonly stipulate rejection of fruit lots with an incidence of internal defects >2%, but current assessment, based on destructive assessment of a few fruit per lot, is unreliable. Non-invasive detection of three defects of commercial importance, apple browning, mandarin dryness and pineapple translucency was attempted using several methods. Reference assessment methods were explored for each defect, including (i) juiciness; (ii) visual assessment of a cut surface; (iii) image analysis of a cut surface. Density sorting, chlorophyll fluorescence and acoustic methods were not consistently associated with the apple disorder. Computed tomography X-ray was capable of imaging the mandarin disorder but line X-ray was also not reliable. Practical results were obtained using a shortwave near infrared transmission unit, employing partial least square regression and linear discriminant analysis for detection of defect in apple and mandarin fruit.