Perceiver – A Five-view Stereo System for High-quality Disparity Generation and its Application in Video Post-Production

It appears your Web browser is not configured to display PDF files. Download adobe Acrobat or click here to download the PDF file.

Click here to download the PDF file.

Supplemental Files: 

Creator: 

Zhu, Chang An

Date: 

2020

Abstract: 

Element extraction from videos has always been a time-consuming process in the entertainment industry. In this research, we explored the possibility of simplifying the video object extraction technique with corresponding depth sequences. Based on post-production quality requirements, we developed our disparity enhancing system by integrating our two-axis-multi-view-stereo method that perceives an environment from five different perspectives on both x and y axes. Our research results have shown that the disparity quality of our approach is both visually and quantitatively more accurate than the traditional one-stereo-pair method, and its object extraction (i.e., matting) quality is comparable with existing mature matting technique to a certain extent. This research output can be applied in video object cut-out, visual effects composition, video's 2D to 3D conversion, and image post-processing. With further improvement, our system might be applicable in AR, VR, machine vision, and auto-pilot areas.

Subject: 

Computer Science

Language: 

English

Publisher: 

Carleton University

Thesis Degree Name: 

Master of Information Technology: 
M.I.T.

Thesis Degree Level: 

Master's

Thesis Degree Discipline: 

Digital Media

Parent Collection: 

Theses and Dissertations

Items in CURVE are protected by copyright, with all rights reserved, unless otherwise indicated. They are made available with permission from the author(s).