Abstract
With our single-wavelength spectral-imaging-based Thai jasmine rice identification system, we emphasize here that a combination of an appropriate polynomial fitting function on the determined chain code and a well-trained neural network configuration is highly sufficient in achieving a low false acceptance rate (FAR) and a low false rejection rate (FRR). Experimental demonstration shows promising results in identifying our desired Thai jasmine rice from six unwanted rice varieties with FAR and FRR values of 6.2% and 7.1%, respectively. Additional key performances include a much faster identification time of 30.5 s, chemical-free analysis, robustness, and adaptive learning.
© 2014 Optical Society of America
Full Article | PDF ArticleMore Like This
Kajpanya Suwansukho, Sarun Sumriddetchkajorn, and Prathan Buranasiri
Appl. Opt. 50(21) 4024-4030 (2011)
Ruoling Deng, Long Qi, Weijie Pan, Zhiqi Wang, Dengbin Fu, and Xiuli Yang
J. Opt. Soc. Am. A 39(6) 1034-1044 (2022)
Kosom Chaitavon, Sarun Sumriddetchkajorn, Anchalee Prasertsak, Sataporn Chanhorm, Panintorn Prempree, and Yuttana Intaravanne
Appl. Opt. 61(8) E14-E20 (2022)