Assessment of citrus fruit quality using a real-time machine vision system

Spain is a major producer of citrus fruit, which must be correctly classified depending on its external quality. Current commercial sorters based on machine vision cannot solve problems like defect detection or correct colour classification due to the low image resolution, which is necessary to achieve adequate production speed. This paper describes a new machine vision system to classify different species of citrus in real time, attending to external quality features of the fruits as size, colour and external defects. A specific hardware has been developed to run the algorithms in parallel, which makes possible work at speed of 10 fruits/s, with an adequate image resolution.