This page gives you an example of how a disparity map between
two camera images can be used to restore some of the
3D information about the real world. Given two real camera images,
we first estimate the disparity map using a disparity space based
technique. An intermediate view is then generated by
linear interpolation from the left and the right camera images.
The occluded regions are interpolated based on a fronto-parallel model,
that is, the occluded regions are assumed to be planes parallel to the
camera baseline. When these images are displayed sequentially,
3D depth can be perceived as follows:
|