Method of directed enumeration of alternatives in the problem of automatic recognition of half-tone images

被引：0

作者：

A. V. Savchenko

机构：

[1] State University Higher School of Economy,

来源：

Optoelectronics, Instrumentation and Data Processing | 2009年 / 45卷 / 3期

关键词：

automatic image recognition; image recognition; learning recognition; minimum information discrimination criterion;

D O I：

10.3103/S8756699009030066

中图分类号：

学科分类号：

摘要：

The problem of automatic recognition of half-tone images based on the minimum information discrimination principle is formulated and solved. A method of directed enumeration of the set of alternatives in the Kullback-Leibler information metric is proposed as opposed to the method of complete enumeration of competing hypotheses. A program and results of an experimental study of the method are presented. It is shown that the proposed algorithm is characterized by increased accuracy and reliability of automatic image recognition.

引用

页码：255 / 261

页数：6

共 6 条

[1]

Voskoboinikov Yu. E.(2004)Choosing the Stopping Iteration in Iterative Algorithms of Image and Signal Reconstruction Avtometriya 40 3-10

[2]

Litvinov L. A.(2003)Mathematical Methods for Material Study and Diagnostics Zavod. Lab. 69 53-64

[3]

Orlov A. I.(1997)Discrimination of Random Signals in the Frequency Domain Radiotekh. Elektron. 42 426-431

[4]

Savchenko V. V.(2005)Detection of the Stochastic Process Discord Based on the Principle of Minimum Informative Divergence Avtometriya 41 68-74

[5]

Akatiev D. Yu.(undefined)undefined undefined undefined undefined-undefined

[6]

Savchenko V. V.(undefined)undefined undefined undefined undefined-undefined

← 1 →