3D RECONSTRUCTION OF A SCENE FROM MULTIPLE 2D IMAGES

Computer vision and augmented reality industry focus on reconstructing 3D view of an object for the purpose of improvising the visual effect. There are various applications which are in need of 3D Models like street views, film industry, game industry etc. The 3D Models cannot be directly captured using video or still cameras. The cameras have the capability to acquire high quality 2D images at various angles. Our paper focuses on identifying the feature points from 2D image acquired from calibrated camera and reconstructing those features using depth analysis to obtain the 3D view of the object or image. The final obtained 3D view would be a dense reconstruction of the input 2D Image. This reconstruction methodology is less complex when compared to traditional mathematical modelling techniques. The methodology could be easily incorporated into any application domain like surveillance in the form of facial reconstruction of 3D.