Skip to main content
ARS Home » Northeast Area » Beltsville, Maryland (BARC) » Beltsville Agricultural Research Center » Environmental Microbial & Food Safety Laboratory » Research » Publications at this Location » Publication #270788

Title: Hyperspectral near-infrared reflectance imaging for detection of defect tomatoes

item LEE, HOONSOO - Chungnam National University
item Kim, Moon
item JEONG, DANHEE - Hanyang University
item Chao, Kuanglin - Kevin Chao
item CHO, BYOUNG-KWAN - Chungnam National University
item Delwiche, Stephen - Steve

Submitted to: Proceedings of SPIE
Publication Type: Proceedings
Publication Acceptance Date: 8/5/2011
Publication Date: 8/16/2011
Citation: Lee, H., Kim, M.S., Jeong, D., Chao, K., Cho, B., Delwiche, S.R. 2011. Hyperspectral near-infrared reflectance imaging for detection of defect tomatoes. Proceedings of SPIE. 8027:1-9.

Interpretive Summary:

Technical Abstract: Cuticle cracks on tomatoes are potential sites of pathogenic infection that may cause deleterious consequences both to consumer health and to fresh and fresh-cut produce markets. The feasibility of a hyperspectral near-infrared imaging technique in the spectral range of 1000 nm to 1700 nm was investigated for detecting defects on tomatoes. Spectral information obtained from the regions of interest on both defect areas and sound areas were analyzed to determine an optimal waveband ratio that could be used for further image processing to discriminate defect areas from the sound tomato surfaces. Unsupervised multivariate analysis methods, such as principal component analysis, were also explored to improve detection accuracy. Threshold values for the optimized features were determined using linear discriminant analysis. Results showed that tomatoes with defects could be differentiated from the sound ones with an overall accuracy of 94.4%. The spectral wavebands and image processing algorithms determined in this study could be used for multispectral inspection of defects on tomatoes.