基于小波域HMT模型的图像超分辨率重构

Wavelet-Domain HMT-Based Image Superresolution

小波域HMT模型采用混合高斯分布 ,并通过多尺度小波系数隐状态之间的Markov依赖性刻画自然图像小波系数随尺度减小呈指数衰减的特性 由于小波域HMT准确刻画了自然图像小波变换的统计特性 ,因此文中算法以此作为自然图像的先验模型 ,并把图像超分辨率问题表述为一个约束优化问题 ,采用Cycle Spinning方法抑制重构出的高分辨率图像中可能存在的震铃和锯齿等失真 最后 。

Wavelet domain hidden Markov tree (HMT) models the dependencies of multiscale wavelet coefficients through the state probabilities of the wavelet coefficients, whose distribution densities can be approximated by the Gaussian mixture Because wavelet domain HMT accurately characterizes the statistics of real world images, the presented algorithm specifies the prior distribution of the real world image through wavelet domain HMT Cycle Spinning technique is used to suppress the artifacts that may exist in the reconstructed high resolution images Experimental results show that the algorithm properly retrieves various kinds of edges and the reconstructed images have high PNSR Quantitative error analyses are provided and several images are shown for subjective assessment