Detection of cuticle defects on cherry tomatoes using hyperspectral fluorescence imagery

Cherry tomatoes are one of the major vegetables consumed in the fresh-cut market. However, the quality evaluation process, which is dependent on simple size- or color-sorting techniques, is inadequate to meet increased consumer demands for high quality and safety. Among various quality evaluations, detection of cracking defects in cherry tomatoes is a critical process since this type of damage can harbor pathogenic microbes that may have detrimental consequences on consumer health. In this study, a multi-spectral fluorescence imaging technique has been presented as a diagnostic tool for non-destructive detection of defective cherry tomatoes. Fluorescence intensity in the area of cracked cuticle was significantly higher in the blue-green spectral region than that of the sound surfaces, suggesting the multi-spectral fluorescence imaging technique as an effective classification tool for detecting cracking defects on cherry tomatoes. Simple ANOVA classification analysis and principal component analysis were employed to investigate optimal fluorescence wavebands. The results illustrate that a multi-spectral fluorescence image in linear combination with a pair of selected wavebands based on the results of ANOVA analysis was able to detect defective cherry tomatoes with >99% accuracy. The detection algorithm investigated in this study is expected to be used to develop on-site and real-time multi-spectral systems for quality evaluation of cherry tomatoes in postharvest processing plants.

► The fluorescence intensity of the areas of cuticle cracks was higher than that of the sound surfaces in the blue-green spectral region. ► The fluorescence emissions from the exposed inner layers of the cherry tomatoes were highly variable perhaps owing to different concentrations of fluorophores. ► A multispectral fluorescence image linearly combined with a few selected wavebands based on the result of supervised discrimination analysis was capable of detecting defective cherry tomatoes with above 99% accuracy.