Creating avatar from a photograph

• Main Author: Low, Shao Jing
• Other Authors: Anamitra Makur
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Electrical and electronic engineering
• Online Access: https://hdl.handle.net/10356/157274
• Institution: Nanyang Technological University

The internet has resulted in an increase in number of shared virtual reality spaces, allowing individuals to interact in the virtual world with an avatar, which is an image representation of themselves. Avatars can be created in a variety of methods and majority of the existing games or internet services requires users to manually customise their own avatar character. Researchers have developed a number of approaches for automatically generating avatars, the majority of which use non-photorealistic rendering techniques to simplify images and generate avatars. The objective of an avatar generator is to generate a cartoon representation of an input image. With many cartoon styling tools available, users can browse through and select the preferred avatar, although this can be time consuming. In this project, we will create a personalised avatar system that uses MATLAB image processing algorithms to construct a cartoon-like appearance from any digital image automatically. The six phases will be used to construct an avatar generator: library preparations, facial features detection, facial features extraction, facial features matching, facial features placement, and avatar generation. The generator will be able to do face detection and extract all facial components such as the face, eyes, nose, mouth. The extracted facial components will then be match to the corresponding avatar components. The avatar generator will generate an avatar by combining the components that are the most identical.