Near-IR and color imaging for bruise detection on Golden Delicious apples

Digital images of reflected light in both the near-infrared (NIR) and visible wavelengths from the surface of bruised and unbruised Golden Delicious apples were captured for classifying bruise damage. Each of the attributes of two models for color representation, RGB and HSI, were compared to NIR for their ability to discriminate bruised from unbruised tissue. The surface reflectance for good tissue decreased from the fruit center outward, except saturation which increased. Reflectance of good tissue also varied adjacent to the bruised area compared to a location 60 degrees away. NIR, green, hue, and red were the features which showed the most contrast between bruised and undamaged tissue. This contrast did not decrease for green, red, and hue as storage time increased.