ipProjector: A Vision-Based Framework for Unsupervised Ubiquitous Object Augmentation Using a Portable Projector-Camera System

摘要

ipProjector stands for interactive portable projector which is a project trying to make a smart projection system while retaining absolute portability of the devices and setups. In this paper, an ipProjector framework is proposed for unsupervised real-time object augmentation applications. Scene-independent projector-camera geometric calibration is introduced so that the projected augmentation always appears in the desired geometry despite an unconstrained surface. Using nonintrusive projection, the system can ensure correct vision-based analysis because external appearances of the environment and object (as seen by the camera) are hardly changed by the overlaid projected content. Finally, an unsupervised visual detector is used to guarantee robustness of the detection regardless of lighting variations. Several image processing algorithms are applied in the framework, including image warping using perspective transformation, color conversion using the nearest-neighbor search algorithm, and multiscale visual detection. Based on a small-scale laboratory setting, experiments were conducted to evaluate the proposed approaches, and one example was built to prove the proposed concept in object augmentation.