Inserting Virtual Static Object with Geometry Consistency into Real Video

Abstract

Given the importance of augmented reality in recent years and its wide spread in many fields of media industries, practical fields such as medicine, education, and others. This study present a methodology in order to insert virtual static object into a real video environment for improve the reality with a virtual information that was not actually present. There are several problems needs to be overcome during the process of merging the virtual object such as geometry of shapes, registration, and illumination. The aim of this study is to make the virtual static objects look like it actually exists in the same real video environment in terms of dimensions, sizes and consistency of shapes. The following steps of the proposed method are: Firstly convert original video file into frames. Secondly separate the background from foreground by rebuilding the background model, then obtain the foreground objects by subtract original frames of the video from background. Thirdly apply the background segmentation. Then extracting the features for every one of the segments and extracting the features of virtual static object. Fifthly selecting a suitable segment (area) for insert the virtual object.