FEASIBILITY AND PERFORMANCES OF A NEW, MULTIPLEXED, FAST AND LOW-COSTFIBEROPTIC NIR SPECTROMETER FOR THE ONLINE MEASUREMENT OF SUGAR IN FRUITS

This paper describes an attempt to develop a new NIR spectrometer to s ort fruits according to the sugar content in the conditioning stations . This spectrometer, ranging from 800 to 1050 nm, is based on diode ar ray technology. It is coupled with fiber optics, which allow multiplex ing; 20 fibers have been detected by the camera, showing the potential multiplexing of 20 different points. The acquisition time is 40 ms pe r image (i.e., 2 ms per spectrum). A good signal-to-noise ratio has be en recorded in the 820-1010 nm range. This spectrometer has been teste d with a set of 79 peaches. Partial least-squares processing is applie d to different wavelength ranges. The best coefficient of correlation, achieved with a reduced wavelength range (from 847 to 977 nm), is 0.8 1, and the standard error of prediction is 1.04-degrees-Brix. With thi s calibration, the fruits can be sorted into three maturity classes wi th a reliability of 76%. This performance, as well as the other charac teristics of the instrument, is in accordance with the requirements of the conditioning station managers.