当前位置: X-MOL 学术Opt. Mem. Neural Networks › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Interpolation of Multidimensional Signals Based on Optimization of Entropy of Postinterpolation Remainders
Optical Memory and Neural Networks ( IF 1.0 ) Pub Date : 2019-02-01 , DOI: 10.3103/s1060992x18040069
M. V. Gashnikov

Abstract

We analyze adaptive algorithms for interpolation of multidimensional signals and propose interpolators based on auto-switching between simple interpolation functions in each point of a signal depending on local characteristics of the signal at this point. The switching is performed with the aid of a parametrized decision rule. To fined optimal values of the parameters of this decision rule, we use a criterion of the minimum energy of postinterpolation remainders and a criterion of the minimal entropy of quantized postinterpolation remainders. We discuss results of application of the proposed interpolator for solving the problems of superimposition of heterogeneous signals and compression of signals. We used real signals and performed computer simulations, which allowed us to compare the proposed interpolators with their prototypes and to estimate the gain resulting from their implementation.


中文翻译:

基于后插值余量熵优化的多维信号插值

摘要

我们分析了用于多维信号插值的自适应算法,并提出了一种插值器,该插值器基于信号在每个点上的简单插值函数之间的自动切换(取决于该点信号的局部特性)而提出。借助参数化的决策规则执行切换。为了细化该决策规则的参数的最优值,我们使用了后插值余量的最小能量的准则和量化后插值余数的最小熵的准则。我们讨论了所提出的插值器用于解决异构信号叠加和信号压缩问题的应用结果。我们使用了真实信号并进行了计算机模拟,
更新日期:2019-02-01
down
wechat
bug