Determination of Intrinsic Qualities of Rice Using Near-Infrared Imaging and Light-Emitting Diodes Detecting Systems
Date Issued
2007
Date
2007
Author(s)
Lin, Lian-Hsiung
DOI
zh-TW
Abstract
One of the objectives of this research was to develop a near-infrared (NIR) imaging system that would detect rice intrinsic qualities nondestructively in real time rice processing lines. The sorting of a single rice kernel based on intrinsic qualities will precisely influence the classification and packaging process of rice. Therefore, the other objective of the research was to develop a rice moisture detecting system for single rice kernel.
The developed NIR imaging system consists mainly of a NIR CCD camera which is coupled to a camera controller. A frame grabber board was used to receive the video signal from the camera. A filter exchange device consisted of a filter adapter, a filter holder, and a stepper motor module that was combined with the CCD camera. The filters were installed in a filter holder. The filter exchange device was controlled by a stepper motor in order to rotate automatically such that the NIR imaging system can effectively acquire multi-spectral images. In this work, calibration methods including multiple linear regression (MLR), partial least squares regression (PLSR), and artificial neural network (ANN) were used in both of near-infrared spectrometer (NIRS) and NIR imaging system to determine the moisture and protein contents of rice. Comprehensive performance comparisons among MLR, PLSR, and ANN approaches were conducted. To reduce repetition and redundancy in the input data and obtain a more accurate network, six and five significant wavelengths selected by the MLR model, which had high correlation with the moisture and protein contents of rice, were used as the input data of the ANN.
The performance of the developed system was evaluated via a series of experimental tests for rice moisture and protein contents. Utilizing three models of MLR, PLSR, and ANN, the rice moisture analysis results of rval2, SEP, and RPD for the validation set were within 0.942-0.952, 0.435-0.479%, and 4.2-4.6, respectively. The prediction of protein content with the NIR imaging system by employing the same three models achieved rval2 of 0.769-0.806, and SEP of 0.266-0.294%, respectively. While compared with a commercial NIRS, experimental results showed that the performance of the NIR imaging system was almost the same as that of NIRS. Using the developed NIR imaging system, all of the three different calibration methods (MLR, PLSR, and ANN) provided satisfactory prediction of rice moisture and protein content. These results indicated that the NIR imaging system developed in this research can be used as a device for the measurement of rice moisture and protein content.
A NIR light-emitting diode (LED) individual rice kernel moisture content measurement system contains NIR LED, rice moving chute, detecting units, and signal processing unit was also developed in this research. A calibration set which contained 78 rice kernels with moisture content ranged from 10.34-22.37% was used to calibrate the system and to develop a prediction equation. Another set of rice which containing 60 kernels with moisture content range of 10.50-21.65% was used for validation. The coefficients of determination for the calibration and validation sets based on 940, 1,050 and 850 nm LED were 0.706 and 0.624, respectively. The results indicated that the developed NIR LED measurement system can be utilized on the rice processing line.
Subjects
近紅外線
影像
含水率
蛋白質
發光二極體
稻米
Near-infrared
Imaging
Moisture content
Protein content
Light- emitting diode
Rice
Type
thesis
File(s)![Thumbnail Image]()
Loading...
Name
ntu-96-D90631003-1.pdf
Size
23.31 KB
Format
Adobe PDF
Checksum
(MD5):9113584ecdc19301166fb000b7b1a453