A novel improved CNN algorithm via denoising approach

Jie Zhang; Jingjing Liu; Mingyu Wang; Aiying Guo; Yuyao Xiao

doi:10.1117/12.2550362

27 November 2019 A novel improved CNN algorithm via denoising approach

Jie Zhang, Jingjing Liu, Mingyu Wang, Aiying Guo, Yuyao Xiao

Proceedings Volume 11321, 2019 International Conference on Image and Video Processing, and Artificial Intelligence; 113212G (2019) https://doi.org/10.1117/12.2550362
Event: The Second International Conference on Image, Video Processing and Artifical Intelligence, 2019, Shanghai, China

Abstract

Address-event-based Dynamic Vision Sensor(DVS) and Convolutional Neural Network(CNN) have been widely researched in recent years. However, the collected data of DVS are easily affected by some noise, which makes it difficult to identify the target during the classification processing. In order to solve the problem of misclassification, a novel improved CNN(NI-CNN) technique is proposed in this paper. Firstly, the appropriate number of event pulses are chosen and mapped to the frame domain, then the optimization denosing approach is utilized to the whole classification system. Secondly, reducing intra-class spacing and enlarging inter-class divergence by joint loss function which is adjusted regularization parameters. Numerical comparisons between our proposed approach and some state-of-the-art solvers, on several accessible databases, are presented to demonstrate its efficiency and effectiveness.

Citation Download Citation

Jie Zhang, Jingjing Liu, Mingyu Wang, Aiying Guo, and Yuyao Xiao "A novel improved CNN algorithm via denoising approach", Proc. SPIE 11321, 2019 International Conference on Image and Video Processing, and Artificial Intelligence, 113212G (27 November 2019); https://doi.org/10.1117/12.2550362

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available