A video inpainting method based on deep flow estimation

Ziji Liu; Huiying Jia; Jiaqi Yang

doi:10.1117/12.3033019

5 July 2024 A video inpainting method based on deep flow estimation

Ziji Liu, Huiying Jia, Jiaqi Yang

Proceedings Volume 13184, Third International Conference on Electronic Information Engineering and Data Processing (EIEDP 2024); 131841P (2024) https://doi.org/10.1117/12.3033019
Event: 3rd International Conference on Electronic Information Engineering and Data Processing (EIEDP 2024), 2024, Kuala Lumpur, Malaysia

Abstract

With the increasing popularity of digital video applications, video restoration techniques have become increasingly important. This paper presents a flow-based video restoration method that aims to achieve high-quality video restoration by analyzing the spatio-temporal relationships between video frames. Specifically, we employ two key steps: flow preprocessing and inter-frame restoration network. In the flow preprocessing stage, frame differencing and optical flow estimation are utilized to obtain the appearance and motion features between frames. The inter-frame restoration network learns feature representations and restoration capabilities by compensating for flow estimation errors during the frame alignment process. We conducted extensive experimental evaluations on multiple video datasets. The experimental results demonstrate significant improvements in restoration quality, spatio-temporal coherence, and real-time performance.

(2024) Published by SPIE. Downloading of the abstract is permitted for personal use only.

Citation Download Citation

Ziji Liu, Huiying Jia, and Jiaqi Yang "A video inpainting method based on deep flow estimation", Proc. SPIE 13184, Third International Conference on Electronic Information Engineering and Data Processing (EIEDP 2024), 131841P (5 July 2024); https://doi.org/10.1117/12.3033019

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