Based Handheld Devices, Augmented Reality Research And Applications

Augmented Reality(AR) technology deals with the combination of real-world and computer-generated data,where computer graphics objects are blended into real footage in real-time.AR technology applies in many fields,including military,medical treatment,education,culture and entertainment.As the increasing development of mobile technology,small and low power handheld devices,such as smart phones and PDAs,have become an ideal platform for AR.Therefore,it is meaningful for researching AR system on handheld devices.Tracking is a most important technology in AR system.In this paper,marker tracking technology based on smart phones shipped with Symbian OS is researched. By setting a marker in real scene,camera pose according to this marker is calculated and then virtual objects could be attached on it.The main work in this paper is as followed:Firstly,a fast algorithm of detecting comers of marker is proposed.Comers detecting is the first step for recognizing marker and calculating camera pose.By analyzing the popular methods used in comers detecting in AR system,an algorithm based on region growing for finding comers of connected regions is proposed.The experiments show that this algorithm is faster than other existed methods used aimed at finding marker's comers.Therefore it is feasible for end-users' devices with low computational resource.Secondly,a solution for avoiding decreasing the performance of real time of AR system caused by handheld devices' joggling is presented.When user holds a phone with one hand,camera's pose is not always fixed but changes a little.For end-users' devices with low computational resource,when camera joggling,it would decrease the performance of real time of AR system if calculating the camera's pose.Therefore,an algorithm based on window estimating camera's pose is presented for this problem. The experiments show that this algorithm could improve the performance of real time greatly by setting suitable window size.Thirdly,a model parser is presented for the aim of reusing models constructed in ordinary software.Fourthly,an AR game,Bear Hitting,based on Symbian OS phone is developed to testify the validity of the above algorithms.The phone will be joggling because of pressing the keys.With algorithms proposed in this paper it will improve the performance of real time of AR game.