Video matting

• Main Author: Garodia, Chawi
• Other Authors: Deepu Rajan
• Format: Final Year Project
• Language: English
• Published: 2014
• Subjects: DRNTU::Engineering
• Online Access: http://hdl.handle.net/10356/59218
• Institution: Nanyang Technological University

This project is an implementation of Video Matting. The objective of this report is to give the reader a detailed description of the author’s work and learning experience during completion of the Final Year Project at School of Computer Engineering at Nanyang Technological University, Singapore. Matting refers to the problem of accurate foreground estimation in images and videos and is used extensively in image and video editing applications. With the recent advancements in digital cameras, the use of matting techniques for editing or to create interesting and novel composites has gained increasing interest. Consequently, a number of matting techniques have been proposed which efficiently separate the foreground from the background and provide good quality matte from images. An image can be separated into background and foreground accurately using image matting. Thus, in video matting, the dynamic foreground isolation from the foreground is carried out in a video sequence. Video matting is a difficult task and there a lot of problems that it faces and hence the main objective is to make a clear distinction between the background and the foreground such as the foreground can be extracted and pasted into another background and the same to be carried out for the whole video sequence. Many algorithms exist to carry out image and video matting. One such algorithm called ‘compressive sensing’ that performs image matting has been implemented in this project. The aim of the project is to extend the image-matting algorithm to video and derive the results and potential problems. The method has been implemented in MATLAB and it aims to minimize user interaction. This report describes the implementation of the algorithm and testing of the results of the experiment as well as an analysis of the results. Finally, it also attempts to provide a conclusion and a recommendation for the improvement of the algorithm.