Abstract:
In computer vision, the process of 3D reconstruction consists of rebuilding a 3D scene
from a set of images. The frames (images) can be extracted from a video sequence
to reconstruct the captured scene. This research study concentrates on nding the
suitable frames for reconstruction and discarding the rest of the frames through the
process of key frame extraction. The selection process based on the properties of the
frames avoids degenerate cases, helps to improve performance, and enables the 3D
reconstruction process to minimize compute time. Besides key frame extraction, I
developed a browser-based 3D video player to allow users to view the captured video
in three dimensions.