A FAST AND ROBUST MAP-BASED SUPERRESOLUTION RECONSTRUCTION METHOD USING CDHS AND BLOCK-WISE MOTION VECTOR SELECTION

作者：Chen Hsi Kuan^*; Chen Chin Hsing; Wang Shih Jen

来源：International Journal of Innovative Computing Information and Control, 2010, 6(7): 3105-3120.

• Occlusion
• Superresolution (SR)
• Cross-diamond-hexagonal search (CDHS)
• Block-wise motion vector selection (BWMVS)

High computational complexity and occlusion including the extreme case of a scene change are the two most critical problems in superresolution (SR) reconstruction. In this paper, we adopted the Cross-Diamond-Hexagonal Search (CDHS) block-matching algorithm (BMA) to reduce the complexity significantly. The CDHS BMA reduced about 2/3 computation time without sacrificing the reconstruction quality comparable to the full search BMA but it has high noises, unqualified motion vectors, resulted from occlusion existing in each frame of a sequence. We farther proposed a simple and robust block-wise motion vector selection (BWMVS) strategy that can remove occlusion effectively. The stability due to BWMVS was verified by the reconstruction results using the interpolation methods (nearest, bilinear, and bicubic). The experimental results showed that by combining the CDHS BMA and BWMVS, high quality images in SR reconstruction can be obtained without complex computation.

• 出版日期2010-7