Real-time color correction for a photo-based augmented reality system

Abstract

This report proposes methods for correcting the color tone of photographic images in a photo-based augmented reality (AR) system in real time to suit the user's environment. Photo-based AR systems achieve AR by selecting a specific photographic image, which was shot in advance, to suit user's camera position and “pasting” it onto the camera view. Unlike conventional systems, this system does not require a 3D model, so no model rendering, which reduces the rendering load, is necessary. The approach should be effective in terminals with small battery capacities such as smartphones. However, this system becomes problematic if the photograph of the object selected does not fit well with the background lighting conditions when the photograph was shot, because the AR may vary. Thus, the proposed method facilitates more natural AR by correcting the color tone of the photographic image.