Mechanization of the selective harvest of broccoli

TR Number



Journal Title

Journal ISSN

Volume Title


Virginia Polytechnic Institute and State University


An investigation was made of concepts for mechanizing the selective harvest of broccoli. Selective harvesting has advantages over once-over harvesting because of greater yield and reduced handling requirements. Results of a preliminary experiment measuring broccoli stalk cutting forces indicated that the blade speed for a broccoli cutting mechanism should be as fast as possible to minimize the required cutting force. A manually-directed, powered cutting device was designed to fit readily into existing broccoli harvest systems. In tests the first year with the device, the harvest rate was substantially faster than hand harvest rates measured at commercial farms, but the level of leaf-stripping achieved with the device was unacceptably low. A new cutting device included an added leaf-stripping mechanism and had a mounting arrangement that allowed the harvesting of two double rows at once. In tests the second year, leaf-stripping was much improved, but the overall harvest rate was only marginally better because of extra manipulation required to activate the leaf-stripping mechanism.

Measurements related to mechanical harvesting were made on broccoli plants both years. Head height, stalk diameter, and head weight were strongly affected by harvest time and in-row plant spacing. Height and stalk diameter were moderately correlated to head diameter. A regression model for predicting head diameter from height and stalk diameter indicated that potential exists for using a combination of the two parameters for sizing broccoli heads. Head weight was highly correlated to height and stalk diameter.

Two concepts for automatic mature head selection were evaluated. The results of an experiment measuring the force required to uproot broccoli plants indicated that physically sizing broccoli heads using spaced fingers would only be feasible if late season irrigations could be incorporated in a harvest system. Digital image processing for head selection appears more promising. An image processing algorithm based on the gray level run length method of textural analysis was developed for predicting broccoli head area. Accurate head classification was obtained with the model. For an automatic selective harvester, an image processing system can be coupled with a cutting device with the major advantage that leaf~stripping can be accomplished automatically during the harvesting action.