Mean Shift: A Robust Approach Toward Feature Space Analysis
IEEE Transactions on Pattern Analysis and Machine Intelligence
Generating the depth map from the motion information of H.264-encoded 2D video sequence
Journal on Image and Video Processing - Special issue on selected papers from multimedia modeling conference 2009
Hi-index | 0.00 |
A 2D to 3D video conversion scheme based on color segmentation and motion information is presented in this paper. Optical flow with initial input from MPEG4 is employed to acquire high quality motion information, which can guarantee better depth information; color segmentation in this scheme divides the reference frame into homogeneous regions and restricts the objects boundaries more accurately; based on the divided regions, the depth values from the motion vectors are used to derive a group of depth plane by plane estimation. The experimental results show that our scheme can provide high quality depth map with accurate objects' boundaries.