[1]肖 汉,杜 莹,王 平,等.一种基于GPU计算的自适应局部降噪并行算法[J].南京师大学报(自然科学版),2025,48(04):139-152.[doi:10.3969/j.issn.1001-4616.2025.04.014]
　Xiao Han,Du Ying,Wang Ping,et al.An Adaptive Local Denoising Parallel Algorithm Based on GPU Computing[J].Journal of Nanjing Normal University(Natural Science Edition),2025,48(04):139-152.[doi:10.3969/j.issn.1001-4616.2025.04.014]
点击复制

一种基于GPU计算的自适应局部降噪并行算法()

《南京师大学报（自然科学版）》[ISSN:1001-4616/CN:32-1239/N]

卷:
48
期数:
2025年04期
页码:
139-152
栏目:
计算机科学与技术
出版日期:
2025-08-20

文章信息/Info

Title:
An Adaptive Local Denoising Parallel Algorithm Based on GPU Computing
文章编号:
1001-4616(2025)04-0139-14
作者:
肖 汉1; 3; 杜 莹2; 王 平1; 周清雷3
(1.郑州师范学院信息科学与技术学院,河南 郑州 450044)
(2.郑州师范学院地理与旅游学院,河南 郑州 450044)
(3.郑州大学计算机与人工智能学院,河南 郑州 450000)
Author(s):
Xiao Han1; 3; Du Ying2; Wang Ping1; Zhou Qinglei3
(1.School of Information Science and Technology,Zhengzhou Normal University,Zhengzhou 450044,China)
(2.School of Geography and Tourism,Zhengzhou Normal University,Zhengzhou 450044,China)
(3.School of Computer and Artificial Intelligence,Zhengzhou University,Zhengzhou 450000,China)
关键词:
自适应滤波器; 局部噪声; 信噪比; GPU; CUDA; 并行算法
Keywords:
adaptive filter; local noise; signal-to-noise ratio; GPU; CUDA; parallel algorithm
分类号:
TP311
DOI:
10.3969/j.issn.1001-4616.2025.04.014
文献标志码:
A
摘要:
随着获取图像像幅规模的增大和分辨率的提高,自适应局部降噪算法的性能成为制约图像实时处理的关键. 本文提出了一种基于GPU的自适应局部降噪并行算法. 从向量化访存、数据本地化计算以及资源配置优化3个方面出发,结合算法特性和底层硬件架构特征,研究了自适应局部降噪算法在CPU+GPU异构计算平台上的并行计算和性能优化. 实验结果显示,在处理8 182×8 182分辨率的图像时,相比CPU串行计算获得了27.39倍加速比,具有较好的数据扩展性. 并行算法充分发挥了GPU的并行处理能力. 文中提出的方法对图像处理算法的GPU加速提供了新的研究思路.
Abstract:
With the increase of the size of the acquired image and the improvement of the resolution,the performance of the adaptive local denoising algorithm becomes the key to restrict real-time processing of image. In this paper,a parallel algorithm of adaptive local denoising based on GPU is proposed. From the three aspects of vectorized memory access,data localization computing and resource allocation optimization,combined with the characteristics of the algorithm and the underlying hardware architecture,the parallel computing and performance optimization of adaptive local denoising algorithm on CPU+GPU heterogeneous computing platform are studied. The experimental results show that when processing 8 182×8 182 resolution images,compared with the CPU serial calculation,the speedup ratio is 27.39 times,which has better data scalability. The parallel algorithm gives full play to the parallel processing ability of GPU. The method proposed in this paper provides a new research idea for GPU acceleration of image processing algorithms.

参考文献/References:

[1]HAJIABADI M. Exploring the potential of parallel adaptive filters for audio noise removal[J]. Signal image and video processing,2024,18(1):445-454.
[2]LIU X B,ZHANG Y W,JING H D,et al. One image segmentation method using U-net and res_unet convolutional networks[J]. Rsc advances,2020,10(16):9396-9406.
[3]GAO W F,LIU F,LIU J,et al. Spatial-spectral adaptive learning with pixelwise filtering for hyperspectral image classification[J]. IEEE transactions on geoscience and remote sensing,2023,61(8):1-15.
[4]燕红文,邓雪峰. 中值滤波在数字图像去噪中的应用[J]. 计算机时代,2020,39(2):47-49.
[5]刘智嘉,夏寅辉,杨德振,等. 基于中值滤波器的红外图像噪声处理的改进方法[J]. 激光与红外,2019,49(3):376-380.
[6]黄琪,李鼎权,施涛,等. 一种针对椒盐噪声的新型迭代自适应滤波器[J]. 南开大学学报(自然科学版),2015,48(5):84-89.
[7]薄振桐. 基于FPGA的实时图像自适应中值滤波器设计[J]. 电子科技,2014,27(1):121-124.
[8]SRINIVAS N,SINGH P,PRADHAN G. A novel hardware architecture for non-local means adaptive filter[J]. National academy science letters:India,2020,62(2):1-4.
[9]KUMAR S,JHA R K. An FPGA-based design for a real-time image denoising using approximated fractional integrator[J]. Multidimensional systems and signal processing,2020,26(2):1-23.
[10]MAIER A,WIGSTROM L,HOFMANN H G,et al. Three-dimensional anisotropic adaptive filtering of projection data for noise reduction in cone beam CT[J]. Medical physics,2011,38(11):5896-5909.
[11]WANG Q Y,CHEN D S,LI S R,et al. An adaptive cartoon-like stylization for color video in real time[J]. Multimedia tools and applications,2017,76(15):16767-16782.
[12]DORONIN A,MEGLINSKI I. Imaging of subcutaneous microcirculation vascular network by double correlation optical coherence tomography[J]. Laser and photonics reviews,2013,7(5):797-800.
[13]RUBIO C,GONZALO J,SIMINSKI J,et al. Efficient computation of the geopotential gradient in graphic processing units[J]. Advances in space research,2024,74(1):332-347.
[14]CHOWTHRI V,UMA A,KALPANA P. Design and implementation of low complexity LMS adaptive filter[J]. International journal of nonlinear analysis and applications,2021,12(SI):1827-1833.
[15]ENGINOGLU S,ERKAN U,MEMIS S. Pixel similarity-based adaptive riesz mean filter for salt-and-pepper noise removal[J]. Multimedia tools and applications,2019,78(24):35401-35418.
[16]REN Z L,ZHANG X R,HU Z J,et al. Reducing radiation dose and improving image quality in CT portal venography using 80 kV and adaptive statistical iterative reconstruction-V in slender patients[J]. Academic radiology,2020,27(2):233-243.
[17]KARCZEWICZ M,HU N,TAQUET J,et al. VVC in-loop filters[J]. IEEE transactions on circuits and systems for video technology,2021,31(10):3907-3925.

备注/Memo

备注/Memo:
收稿日期:2024-08-20.
基金项目:国家自然科学基金项目(61572444、61250007).
通讯作者:肖汉,博士,教授,研究方向:并行算法研究与设计、遥感大数据并行处理的研究. E-mail:xiaohan@163.com
更新日期/Last Update: 2025-08-20