An on-line nuclear magnetic resonance (NMR) quality evaluation sensor was d
esigned, constructed and tested. The device consists of a superconducting m
agnet with a 20 mm diameter surface coil and a 150 mm diameter imaging coil
coupled to a conveyor system. The conveyor was run at speeds ranging from
0 to 250 mm/s. The NMR spectra of avocado fruits and one-dimensional magnet
ic resonance images of fresh cherries were acquired while the fruits were m
oving on a conveyor bet. The NMR spectra were used to measure the oil/water
ratio in avocados and this ratio correlated to percent dry weight, the cor
relation coefficient varied between 0.97 and 0.89 and decreased with increa
sing speed. One-dimensional magnetic resonance images of cherries were used
to detect the presence of pits in cherries. Different belt speeds (up to 2
50 mm/s) were used and results of the pit detection tests were compared wit
h static measurements. Higher classification errors occurred under static c
onditions as compared to dynamic conditions. An algorithm based on the chan
ge in shape of the one-dimensional image between a cherry with and without
a pit yielded good classification of the fruits under static and dynamic co
nditions. (C) 1999 Silsoe Research Institute.