Scalable Multi-projector Augmented Reality System

Augmented reality is the technology to increase user’s perception ofthe real world through information provided by the computer system, Itadds computer-generated virtual objects, scenes and system message ontothe real scene thus achieve augmentation to reality. Augmented realitytechnology is widely used in accurate positioning of medical surgery,military reconnaissance, monument restoration, business shows, gamesentertainment and other fields.Among augmented reality technology, multi-projector augmentedreality system has several unique advantages: it can achieve true seamlessdisplay, it is more flexible, it has cheaper hardware configuration, andwith the rapid development of the instrument manufacturing process ofprojectors it is gradually support higher resolution. With the aboveadvantages, multi-projector augmented reality technology has been morewidely used in augmented reality field.The main work of this paper includes four aspects:The first is the study of multi-projector augmented reality system’svisual defects. In this paper, four types of visual defects, which are geometric distortion, photometric distortion, color distortion and edgedistortion, are studied. Their causes, the calibration ideas addressing them,and the associated calibration algorithm are explained.The second is the introduction of a set of multi-projector imagecompensation algorithm. In this paper, on the basis of summarizing therelated work, three types of visual defects, which are geometric distortion,photometric distortion and edge distortion, are addressed. And thecorresponding correction algorithms are discussed and drawn. Finallythese algorithms are integrated together to get a complete set ofmulti-projector image compensation algorithm.The third is the study of distributed parallel rendering technology formulti-projector augmented reality system. In this paper, distributedparallel rendering algorithms is studied. Common distributed parallelrendering platform, such as WireGL, Chromium and Equalizer, arestudied by introducing their architecture and giving a comparativeanalysis on their strengths and weaknesses.The fourth is the implementation of multi-projector imagecompensation algorithm on parallel rendering platform Equalizer. Thesystem’s architecture and running flow are introduced, the experimentresults are presented, the performance, flexibility and scalability of thesystem are analyzed.The innovations of this paper mainly include two aspects: The first is the introduction of a set of multi-projector imagecompensation algorithm. The algorithm can eliminate three types ofvisual defects, which are geometric distortion, photometric distortion andedge distortion, while keep good extensibility.The second is the implementation of multi-projector imagecompensation algorithm on parallel rendering platform Equalizer. Thesystem has good performance, flexibility and scalability.