A Feasibility Study Using Simplified near Infrared Imaging to Detect Fruit Fly Larvae in Intact Fruit

• Published in: Journal of near infrared spectroscopy (United Kingdom) 2011-02, Vol.19 (1), p.55-60
• Main Authors: Saranwong, Sirinnapa, Haff, Ronald P., Thanapase, Warunee, Janhiran, Athit, Kasemsumran, Sumaporn, Kawano, Sumio
• Format: Article
• Language: English
• ISSN: 0967-0335; 1751-6552
• DOI: 10.1255/jnirs.915

The potential for near infrared (NIR) imaging technology to yield improved sensitivity for detecting oriental fruit flies infested in intact mangoes was investigated. Hyperspectral data from 400 nm to 1000 nm were acquired from eight infested mangoes and eight control mangoes at Oh, 24 h and 48 h after infestation. Sixteen pores were created on each mango. Four mangoes from each class were set aside as independent test samples. From each of the eight remaining fruit, nine spectra (3 × 3 pixel) from the area of each pore were extracted. Each spectrum was considered to be an individual, multiple-feature sample. The features were input into an iterative Bayesian discriminant analysis routine for the classification. Using the three selected wavelengths for spectra measured at 48 h after infestation, classification results were 0.9% false negatives (infested fruit misclassified) (range of fruit fly larvae: 8–57) and 5.7% false positives (control fruit misclassified). Grey-scale images of the area (4 × 4 cm) containing the pores for each fruit were generated based on the Mahalanobis distance calculated from the pretreated 48 h absorbance values at each of the three wavelengths selected. Clear distinction between infested areas were observed with few false positive results in the control fruit, indicating the feasibility of using NIR imaging as the basis for a low-cost, high-speed device for the detection of the oriental fruit fly in mangoes.