Illumination Estimation Based on Human Visual Characteristics for Augmented Reality

Abstract

In AR(Augmented Reality) technology, it is important to synthesize realistic images that users do not feel strange. It is necessary in order to realize photometric registration for synthesizing realistic AR images. It should be estimated illumination environment in the real world. In this paper, questionnaire is taken from users to analyze factors that affect the users to feel strange with synthetic AR images. This paper proposes a simple method based on the human visual characteristics. A light source position estimation method is proposed by using a monocular camera and a cast shadow projected from a cylindrical object. AR images are generated using an estimated light source information. Usefulness of the proposed method is presented by the user questionnaire.