ECSE Departmental Research Seminar, 9th November 2005

Abstract:

When produce is prepackaged and sold by weigh it is often necessary to weigh each individual item. Typical packing lines operate at 10-15 items per second making conventional weighing methods inaccurate. A system of mirrors is used to obtain two perpendicular views of each item. Each view is then segmented from the complex background, and the volume estimated. The system is calibrated dynamically to ensure that the weight is accurate even as the product density varies during the season. The produce is then graded by weight, and a system of actuators removes the produce from the conveyer and drops it down a particular chute to form a bundle. The actuators use distributed control, with dedicated micro-controllers controlling each chute under the control of a PC in a master / slave RS485 network. Each actuator includes a sensor for synchronisation with the conveyer. A daisy chain scheme is used to automatically assign controller addresses on startup. The complete system is operates under the control of LabView running on a standard desktop PC. The image processing system achieves an accuracy of 4.4% per item or 2% per 100 g bundle at rates of up to 30 items per second. The mechanical system is the limiting factor in terms of speed; it is able to process up to 15 items per second. The system is currently in use in a NZ packhouse, packing asparagus for the export market.


About the speaker:

Dr Donald Bailey has a BE(Hons) and PhD in Electrical and Electronic Engineering from University of Canterbury, New Zealand. After spending 2 years applying image analysis techniques to the wool and paper industries within New Zealand, he spent 2½ years as a visiting researcher at the Electrical and Computer Engineering Department at the University of California at Santa Barbara. In 1989, he returned to New Zealand as Director of the Image Analysis Unit at Massey University. The Image Analysis Unit provided a consultancy service for academics and researchers on using image analysis techniques for data gathering. In 1998 he moved to the Institute of Information Sciences and Technology where he is currently senior lecturer and leader of the Image and Signal Processing Research Group, and group leader for the Electronics and InfoComm Engineering academic group. Donald’s research interests include applications of image analysis, machine vision, and robot vision. One area of particular interest is the application of FPGAs to implementing image processing algorithms.