ON THE RECONSTRUCTION ASPECTS OF MOMENT DESCRIPTORS

被引：59

作者：

PAWLAK, M

机构：

[1] Department of Electrical and Computer Engineering, University of Manitoba, Winnipeg, MB

来源：

IEEE TRANSACTIONS ON INFORMATION THEORY | 1992年 / 38卷 / 06期

基金：

加拿大自然科学与工程研究理事会;

关键词：

RECONSTRUCTION FROM MOMENTS; ORTHOGONAL MOMENTS; RECONSTRUCTION ERROR; OPTIMAL NUMBER OF MOMENTS; CROSS-VALIDATION;

D O I：

10.1109/18.165444

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

The problem of reconstruction of an image from discrete and noisy data by the method of moments is examined. The set of orthogonal moments based on Legendre polynomials is employed. A general class of signal-dependent noise models is taken into account. An asymptotic expansion for the global reconstruction error is established. This reveals mutual relationships between a number of moments, the image smoothness, sampling rate and noise model characteristics. The problem of an automatic (data-driven) selection of an optimal number of moments is studied. This is accomplished with the help of cross-validation techniques.

引用

页码：1698 / 1708

页数：11

共 33 条

[1] RECOGNITIVE ASPECTS OF MOMENT INVARIANTS [J].