30 March 2004 Noninvasive maturity detection of citrus with machine vision
Author Affiliations +
Proceedings Volume 5271, Monitoring Food Safety, Agriculture, and Plant Health; (2004) https://doi.org/10.1117/12.516052
Event: Optical Technologies for Industrial, Environmental, and Biological Sensing, 2003, Providence, RI, United States
Abstract
A computer vision system was established to explore a method for citrus maturity detection. The surface color information and the ratio of total soluble solid to titratable acid (TSS/TA) were used as maturity indexes of citrus. The spectral reflectance properties with different color were measured by UV-240 ultraviolet and visible spectrophotometer. The biggest discrepancy of gray levels between citrus pixels and background pixels was in blue component image by image background segmentation. Dynamic threshold method for background segmentation had best result in blue component image. Methods for citrus image color description were studied. The citrus spectral reflectance experiments showed that green surface and saffron surface of citrus were of highest spectral reflectance at the wavelength of 700nm, the difference between them reached to maximum, about 53%, and the image acquired at this wavelength was of more color information for maturity detection. A triple-layer feed forward network was established to map citrus maturity from the hue frequency sequence by the mean of artificial neural network. After training, the network mapper was used to detect the maturity of the test sample set, which was composed of 252 Weizhang citrus with different maturity. The identification accuracy of mature citrus reached 79.1%, that of immature citrus was 63.6%, and the mean identification accuracy was 77.8%. This study suggested that it is feasible to detect citrus maturity non-invasively by using the computer vision system and hue frequency sequence method.
© (2004) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Yibin Ying, Yibin Ying, Zhenggang Xu, Zhenggang Xu, Xiaping Fu, Xiaping Fu, Yande Liu, Yande Liu, } "Noninvasive maturity detection of citrus with machine vision", Proc. SPIE 5271, Monitoring Food Safety, Agriculture, and Plant Health, (30 March 2004); doi: 10.1117/12.516052; https://doi.org/10.1117/12.516052
PROCEEDINGS
11 PAGES


SHARE
Back to Top