Applying the CDHS to Improve the MAP-Based Algorithm in Super Resolution Reconstruction from Sequences
Hsi-Kuan Chen, Chin‐Hsing Chen, Shih-Jen Wang · 2007
Based on the Huber-Markov random field (HMRF), this paper adopts the cross-diamond-hexagonal search (CDHS) blocking-matching algorithm to promote the speed of motion estimation in the maximum a posteriori (MAP)-based super resolution (SR) reconstruction. A block-wised motion vector selection (MVS) strategy is proposed to eliminate the unqualified motion vectors due to occlusion occurred in motion estimation so as to strengthen the reliability of the suboptimal solutions and get a better reconstruction. The experimental results show that the proposed algorithm can save about 2/3 computation time while provides the same reconstruction quality compared to that adopting the full search (FS) blocking-matching algorithm.