Main content area

Detecting decayed peach using a rotating hyperspectral imaging testbed

Sun, Ye, Xiao, Hui, Tu, Sicong, Sun, Ke, Pan, Leiqing, Tu, Kang
Lebensmittel-Wissenschaft + [i.e. und] Technologie 2018 v.87 pp. 326-332
Rhizopus, algorithms, fungi, hyperspectral imagery, monitoring, peaches, statistical analysis, wavelengths
A hyperspectral imaging system with a moving testbed has been developed for detection of the disease caused by Rhizopus stolonifera in peaches. The all-around hyperspectral imaging of the whole peach was obtained, which can identify the decayed area fully and is suitable for online monitoring. Three single-band images (709, 807, and 874 nm) which were selected by statistical methods and an image segmentation algorithm were applied to locate the decayed area of peaches was developed based on band ratio image coupled with a simple thresholding method. The performance of image segmentation algorithm of the single-band images was evaluated. The detection accuracy of peaches classified as ‘sound’, ‘slight-decayed’, ‘moderate-decayed’ and ‘severe-decayed’ were 95%, 66.29%, 100% and 100%, respectively. The spectral information was extracted from the decayed area to improve the detection accuracy. The six optical wavelengths were selected by SPA (successive projections algorithm) from the full spectral range. The classification accuracy of sound and rotten peaches was 100% if only these two categories were applied. Our results demonstrated that the hyperspectral imaging method offers the potential to be used to automatically detect fungal infection in peaches.