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.
A priori information-based lightweight convolutional neural network(CNN)for video denoising is pro⁃posed in this paper.More specifically,the priori information is obtained from the adjacent frames and the pre-denosingbased motion compensation is applied to effectively collect the priori information.Meanwhile,the dual-path processingbased 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.
作者
申屠敏健
朱强
朱树元
孟现东
SHENTU Min-jian;ZHU Qiang;ZHU Shu-yuan;MENG Xian-dong(School of Information and Communication Engineering,University of Electronic Science and Technology of China,Chengdu,Sichuan 611731,China;Peng Cheng Laboratory,Shenzhen,Guangdong 518000,China)
出处
《电子学报》
EI
CAS
CSCD
北大核心
2023年第6期1510-1517,共8页
Acta Electronica Sinica
基金
国家自然科学基金(No.U20A20184)。
关键词
视频去噪
卷积神经网络
先验信息
轻量化网络
特征
video denoising
convolutional neural network
priori information
lightweight network
feature
