Foreign Substance Detection in Blueberry Fruits by Spectral Imaging

抄録

Spectral analysis and image processing of the spectral images of blueberries and several foreign substances clarified that plant organs could be detected by the second derivative absorbance image at 680nm, which is an absorption band of chlorophyll. As a feasibility study, spectral images of blueberries and plant organs placed in the 160×120mm area were acquired at 660, 680 and 700nm in order to develop a second derivative absorbance image at 680nm. Then the second derivative absorbances of 52 blueberries and 26 plant organs in the image and their mean values were calculated. Using these data, the probability of being a plant organ, p, was calculated for each pixel of the second derivative absorbance image in order to develop a plant organ detection image. In the detection image, a pixel with p value larger than 0.95 was judged as a plant organ. The positions of pixels judged as plant organs in the detection image were in good agreement with the actual locations where plant organs had been placed. Therefore, it could be concluded that plant organs contaminated raw blueberry materials could be detected using the method developed in this research.