BROCCOLI HEAD SIZING USING IMAGE TEXTURAL ANALYSIS

ABSTRACT An investigation was conducted to evaluate the feasibility of using a digital image processing system for selecting mature broccoli heads based on size. Images were made under controlled lighting conditions of 48 broccoli plants with a wide range of head sizes. Images were used to develop and test a model based on the gray level run length method of textural analysis. The model exhibited an exponential relationship between head area and the numerical texture measure and had a standard error of head area prediction of 16 cm^, which corresponds to an error of less than 1.0 cm in head diameter estimation for a 10-cm diameter head. The results indicate the model has good capability for classifying broccoli heads into immature and harvestable sizes.