Abstract
Noise insensitivity is one of the most important factors in practical implementations of pattern-recognition systems. Our emphasis is on the recognition of objects in a noisy environment. Spatial filters for distinction between the members of a two-class training set can be designed to achieve minimal error probability in the presence of statistical noise. The filter-design procedure can include additional requirements, such as rotation invariance. Computer simulations demonstrate high class discrimination even in high noise levels that demolish the capability of a human observer to distinguish between the classes.
© 1991 Optical Society of America
Full Article | PDF ArticleMore Like This
Gady Zalman and Joseph Shamir
J. Opt. Soc. Am. A 8(5) 814-821 (1991)
Uri Mahlab and Joseph Shamir
J. Opt. Soc. Am. A 8(8) 1233-1239 (1991)
Richard D. Juday
J. Opt. Soc. Am. A 18(8) 1882-1896 (2001)