Semi-automatic Depth Map Extraction Method for Stereo Video Conversion

摘要

This paper present a semi-automatic method aimed to generate stereoscopic views estimating depth information from a single input video frame and decrease computation resources. Firstly, background modeling divides a 2D video into static background scene and moving object. by the viewer's acquired monocular depth cue, the user-defined points in the static scene are used to extract vanishing line. Then, the background depth map is determined and the moving objects are assigned depth values according to their position in the static scene. Finally, the motion based depth map and the geometry based map are integrated into one depth map by a depth fusion algorithm. Combining the depth map from this work and original 2D video can produce vivid 3D video.