Quantifying aesthetics of visual design applied to automatic design
Abstract
In today's "Instagram" world, with advances in ubiquitous computing and access to social networks, digital media is adopted by art and culture. In this dissertation, we study what makes a good design by investigating mechanisms to bring aesthetics of design from realm of subjection to objection. These mechanisms are a combination of three main approaches: learning theories and principles of design by collaborating with professional designers, mathematically and statistically modeling good designs from large scale datasets, and crowdscourcing to model perceived aesthetics of designs from general public responses. We then apply the knowledge gained in automatic design creation tools to help non-designers in self-publishing, and designers in inspiration and creativity. Arguably, unlike visual arts where the main goals may be abstract, visual design is conceptualized and created to convey a message and communicate with audiences. Therefore, we develop a semantic design mining framework to automatically link the design elements, layout, color, typography, and photos to linguistic concepts. The inferred semantics are applied to a design expert system to leverage user interactions in order to create personalized designs via recommendation algorithms based on the user's preferences.
Degree
Ph.D.
Advisors
Allebach, Purdue University.
Subject Area
Computer Engineering|Electrical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.