Hyperspectral data discrimination methods

David P. Casasent; Xuewen Chen

doi:10.1117/12.411754

29 December 2000 Hyperspectral data discrimination methods

David P. Casasent, Xuewen Chen

Proceedings Volume 4203, Biological Quality and Precision Agriculture II; (2000) https://doi.org/10.1117/12.411754
Event: Environmental and Industrial Sensing, 2000, Boston, MA, United States

Abstract

Hyperspectral data provides spectral response information that provides detailed chemical, moisture, and other description of constituent parts of an item. These new sensor data are useful in USDA product inspection. However, such data introduce problems such as the curse of dimensionality, the need to reduce the number of features used to accommodate realistic small training set sizes, and the need to employ discriminatory features and still achieve good generalization (comparable training and test set performance). Several two-step methods are compared to a new and preferable single-step spectral decomposition algorithm. Initial results on hyperspectral data for good/bad almonds and for good/bad (aflatoxin infested) corn kernels are presented. The hyperspectral application addressed differs greatly from prior USDA work (PLS) in which the level of a specific channel constituent in food was estimated. A validation set (separate from the test set) is used in selecting algorithm parameters. Threshold parameters are varied to select the best P_c operating point. Initial results show that nonlinear features yield improved performance.

Citation Download Citation

David P. Casasent and Xuewen Chen "Hyperspectral data discrimination methods", Proc. SPIE 4203, Biological Quality and Precision Agriculture II, (29 December 2000); https://doi.org/10.1117/12.411754

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available