A Maximally Split and Relaxed ADMM for Constrained Least-Squares Design of Two-Dimensional FIR Filters

MA Meng-yao; LAI Xiao-ping; MENG Hai-long

doi:10.3969/j.issn.0372-2112.2020.03.013

您当前的位置：

首页 >

文章列表页 >

A Maximally Split and Relaxed ADMM for Constrained Least-Squares Design of Two-Dimensional FIR Filters

更新时间：2025-07-16

- A Maximally Split and Relaxed ADMM for Constrained Least-Squares Design of Two-Dimensional FIR Filters
- Acta Electronica Sinica Vol. 48, Issue 3, Pages: 510-517(2020)
- 作者机构：
  
  1. 杭州电子科技大学自动化学院,浙江,杭州,310018
  2. 齐鲁工业大学(山东省科学院)数学与统计学院,山东,济南,250353
  3. 杭州电子科技大学自动化学院,浙江,杭州,310018
  4. 齐鲁工业大学(山东省科学院)数学与统计学院,山东,济南,250353
- 作者简介：
- 基金信息：
  
  National Natural Science Foundation of China (No.61573123, No.U1909209, No.U1509205)
- DOI：10.3969/j.issn.0372-2112.2020.03.013
  CLC： TN79
- Published Online：25 March 2020，
  
  Published：2020
- 稿件说明：
移动端阅览
MA Meng-yao, LAI Xiao-ping, MENG Hai-long. A Maximally Split and Relaxed ADMM for Constrained Least-Squares Design of Two-Dimensional FIR Filters[J]. Acta Electronica Sinica, 2020, 48(3): 510-517.
DOI：

MA Meng-yao, LAI Xiao-ping, MENG Hai-long. A Maximally Split and Relaxed ADMM for Constrained Least-Squares Design of Two-Dimensional FIR Filters[J]. Acta Electronica Sinica, 2020, 48(3): 510-517. DOI： 10.3969/j.issn.0372-2112.2020.03.013.

摘要

约束二维有限脉冲响应（Finite Impulse Response，FIR）滤波器，现有设计算法计算复杂度高.针对二维FIR滤波器的约束最小二乘设计，本文应用交替方向乘子法（Alternating Direction Method of Multipliers，ADMM），研究其并行优化方法.通过模型的最大分划，并采用一种松弛技术，提出一个具有高度并行结构的最大分划松弛ADMM算法，分析了算法的计算复杂度，讨论了算法的收敛性，并给出了算法的参数设置方法.实验表明，最大分划松弛ADMM比非松弛的最大分划ADMM收敛快很多；与现有算法相比，提高了计算效率.GPU加速实验中获得的大加速比，表明了所提算法的高度并行性和可扩展性，在图像处理、计算机视觉、模式识别及机器学习等领域有广阔的应用前景.

Abstract

For constrained two-dimensional (2-D) finite impulse response (FIR) filters

the computational complexity of the existing design algorithms is very high. Based on the alternating direction method of multipliers (ADMM)

the parallel optimization of constrained least-squares (CLS) 2-D FIR filters was studied. By maximally splitting the problem into univariate subproblems and utilizing a relaxation technique

a maximally split and relaxed ADMM with a highly parallel computing architecture was proposed. The computational complexity and convergence of the algorithm were analyzed

and a practical scheme for selection of the algorithm parameters was provided. Experimental results show that the proposed maximally split and relaxed ADMM converges much faster than the maximally split unrelaxed ADMM. Compared with existing algorithms

the computational efficiency of the proposed algorithm is improved. The large acceleration ratios obtained by GPU demonstrate the high parallelism and scalability of the proposed algorithm

which is very valuable for applications of the algorithm in image processing

computer vision

pattern recognition and machine learning.

关键词

Keywords

references

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Learnable Threshold Optimization for Massive Dynamic Multi-User Access Detection

Adaptive Federated Learning Based Crowd Sensing Algorithm for Environmental Monitoring

Local Low-Rank and Bloch Manifold Regularized Magnetic Resonance Fingerprinting Reconstruction

Edge Serial Message Passing Algorithm Based on Subgraph

Generalization of Zero‑Padded Tri‑Mode OFDM with Index Modulation Based on Three‑Stage LLR Detection

Related Author

SHI Chang-wei

GUO Li-ting

KANG Peng

DU Wei-qing

CHEN Ping-ping

FANG Yi

JIANG Wei-jin

DU Xi-chen

Related Institution

School of Information Engineering, Guangdong University of Technology

School of Physics and Information Engineering, Fuzhou University

College of Frontier Intersection, Hunan University of Technology and Business

Hunan University of Information Technology

Xiangjiang Laboratory

⁰