Research On Texture Generation And 3D Object Tracking For Augmented Reality

As smartphones,tablets,and wore a rapid development of display device,aug-mented reality technology obtained widespread attention.Augmented reality has many characteristics,such as the virtual-real fusion and 3D registration.There-fore,realistic rendering of virtual objects and camera position accurate real-time recovery are the two key problems.Centering on the two key problems in aug-mented reality,this paper deeply studies the realistic 3D object texture generation and 3D objects tracking,respectively,puts forward a semiautomatic method for generating high quality texture and a 3D object tracking method,effectively im-proves the user experience of augmented reality applications.In general,the main contributions of this paper are as follows:1)For generating realistic texture,it needs to solve the problem of how to generate more lifelike texture information,in order to ensure consistency texture on the surface of photometric measure,however,the current mainstream of automatic texture generation approach will inevitably produce the photometric distortion.To this end,this paper puts forward a set of semi-automatic method for generating high quality texture,the method includes automatic texture mapping and texture repairing,through the optimization of existing texture mapping algorithm and adding of manual post-processing process,so as to generate more realistic texture information.2)For the 3D object tracking,it needs to solve the problem is how to solve the camera pose stably,the current mainstream of tracking algorithms are based on feature and texture information,however,in real life,part of the object itself has not too much feature,even also has high reflective surface,the current feature tracking method cannot deal with these problems well.To this end,this paper proposes 3D object tracking algorithm for lack of texture information,and inte-grates with the existing SLAM system,so as to track stably for known 3D objects in the strange scene.Experiment and comparison results show that the proposed semi-automatic texture generation method can effectively improve the quality of texture mapping,at the same time,the proposed fusion tracking algorithm of 3D object works well,both of them can satisfy the application requirements of augmented reality.