ASABE Technical Library - Abstract

Abstract. Blueberries are a high-value crop for the US agricultural industry. Bruising of the berries during harvesting and processing is a significant source of monetary losses for the industry. The goal of this study was to investigate the potential of hyperspectral imaging using the transmittance mode to detect and grade blueberry bruises. A total of 90 fruits were evenly divided into three groups: control, stem bruise, and equatorial bruise. Fruits in the bruised treatment groups were dropped from 30 cm onto a steel plate using a pendulum to control the location of impact, while the control group was kept intact. A near-infrared hyperspectral imaging system was implemented to acquire transmittance images at a range of 950 nm to 1400 nm. Images were acquired from three positions for fruit in the control and stem bruise groups, and from four positions in the equatorial bruise group. Color images of sliced fruit were obtained as references. By comparing against the reference color images, the profiles of spatial and spectral intensity were evaluated to determine which wavelengths and imaging positions can most effectively detect bruising. Images at 1070 nm were processed using thresholding and morphological reconstruction to calculate bruise ratios by dividing the number of bruised pixels by the total number of fruit pixels. Correlation analysis was performed between the calculated bruise ratios and human grading for three treatments. The results indicated that it is feasible to use hyperspectral transmittance images to detect and quantify blueberry bruises.