电子学报

• •    

基于视频先验信息的轻量化去噪卷积神经网络

申屠敏健1, 朱强1, 朱树元1, 孟现东2   

  1. 1.电子科技大学信息与通信工程学院,四川成都 611731
    2.鹏城实验室,广东深圳 518000
  • 收稿日期:2021-05-26 修回日期:2022-05-31 出版日期:2023-02-15
    • 作者简介:
    • 申屠敏健 男,1995年9月出生,浙江省东阳人.电子科技大学信息与通信工程学院硕士研究生.主要研究方向为深度学习与视频去噪.E-mail: shentu_mj@163.com
      朱强 男,1997年2月出生,甘肃省定西人.电子科技大学信息与通信工程学院博士研究生.主要研究方向为图像及视频超分辨率.E-mail: zhuqiang@std.uestc.edu.cn
      朱树元 男 1980年1月出生,河南省平顶山人.电子科技大学信息与通信工程学院教授.主要从事图像视频处理、图像视频编码等方面的研究.E-mail: eezsy@uestc.edu.cn
    • 基金资助:
    • 国家自然科学基金(U20A20184)

A Priori Information-Based Lightweight Convolutional Neural Network for Video Denoising

SHENTU Min-jian1, ZHU Qiang1, ZHU Shu-yuan1, MENG Xian-dong2   

  1. 1.School of Information and Communication Engineering,University of Electronic Science and Technology of China,Chengdu,Sichuan 611731,China
    2.Peng Cheng Laboratory,Shenzhen,Guangdong 518000,China
  • Received:2021-05-26 Revised:2022-05-31 Online:2023-02-15
    • Supported by:
    • Foundation Item(s): National Natural Science Foundation of China(U20A20184)

摘要:

提出了一种基于视频先验信息的轻量化去噪卷积神经网络.先验信息从近邻的多帧视频图像中获取,采用了基于预去噪的视频运动补偿方法消除噪声和运动偏移对信息获取准确度的影响.为降低卷积神经网络复杂度,构建了基于双路处理的卷积神经网络用于去除视频噪声,特别是设计了双路稠密连接单元,实现了网络的轻量化.双路稠密连接单元通过高、低分辨率特征分解和特征拼接,有效降低了网络复杂度.实验结果表明:采用本文方法去除视频噪声能够获得较好的客观评价结果和主观视觉结果.此外,在减少网络参数、降低浮点运算次数和提升运行速度方面均体现出了良好性能.

关键词: 视频去噪, 卷积神经网络, 先验信息, 轻量化网络, 特征

Abstract:

A priori information-based lightweight convolutional neural network (CNN) for video denoising is proposed in this paper. More specifically, the priori information is obtained from the adjacent frames and the pre-denosing-based motion compensation is applied to effectively collect the priori information. Meanwhile, the dual-path processing-based CNN is designed to remove the video noise. Specifically, the dual-path cross connection unit (DCU) is proposed to simplify the feature extraction for the design of a lightweight network. With DCU, the high-resolution features and the low-resolution features are generated and they are concatenated by shorter connections, which achieve low complexity. The experimental results demonstrate that our proposed method offers good objective results as well as subjective results. Moreover, it effectively reduces network parameters and floating point operations, and achieves a fast CNN-based video denosing.

Key words: video denoising, convolutional neural network, priori information, lightweight network, feature

中图分类号: