Journal of Software Engineering1819-43112152-0941orgz10.3923/jse.2015.520.533ZhaiHaitianLiHuiSunMingui3201593Super-resolution is the process of combining a sequence of low-resolution images
to produce a higher resolution image. The conventional super-resolution algorithms
usually apply the same regularization factor for the whole image, regardless
of region characteristics. One regularization factor is not good enough for
all regions since an image consists of various regions having different characteristics.
In this study, we propose a region adaptive super-resolution method to apply
an adaptive regularization factor for each separate region. The regions are
generated by segmenting the reference frame using the improved hierarchical
segmentation algorithm. Regularization parameters are then adaptively determined
based on the region characteristics. The software of the proposed algorithm
is also implemented on an Intel quad core computer. Finally, the experimental
results using both synthetic and unsynthetic image sequences show the effectiveness
of the proposed algorithm compared to three state-of-the-art super resolution
algorithms.]]>Park, S.C., M.K. Park and M.G. Kang,2003Song, H., X. He, W. Chen and Y. Sun,2010Li, H. and K.M. Lam,2013Aguena, M.L.S. and N.D.A. Mascarenhas,2006Su, X. and S. Li,2011Panda, S.S., M.S.R.S. Prasad and G. Jena,2011Kasahara, R., T. Ogata, T. Kawasaki, H. Miura and K. Yokoi,2007Chantas, G.K., N.P. Galatsanos and N.A. Woods,2007Belekos, S.P., N.P. Galatsanos and A.K. Katsaggelos,2010Wang, S., S.H. Kim, Y. Liu, H.K. Ryu and H.M. Cho,2010Ng, M.K., H. Shen, E.Y. Lam and L. Zhang,2007Dobigeon, N., J.Y. Tourneret and M. Davy,2007Huang, L.L., L. Xiao and Z.H. Wei,2011