An algorithm to select the most representative 3D
key-frames from 3D video sequences is proposed in this paper.
Optimal 3D key-frames are found in temporal segments of
arbitrary length using a dynamic programming algorithm which
minimises the dissimilarity between the reconstructed and the
original temporal segment. The dissimilarity measure is based
on a colour and depth measure which combines the contribution
from colour and depth perception provide a 3D quality measure
of the 3D summary. The results evaluated using the Shot
Reconstruction Degree (SRD) and the Fidelity measure show that
the key-frame selection process presented here outperforms those obtained from uniform sampling methods