摘要： 在L1范数图像超分辨率重建框架下，引入参数自适应估计，该方法对模型误差表现出良好的稳健性并且可以加速收敛。结合差分图像统计特性和概率先验模型，解释了L1范数形式的双边全变差正则项概念，利用Kullback-Leibler距离证明了该正则项的优越性，并分析了混合先验模型在超分辨率重建中应用的可行性等问题。

关键词: 超分辨率, 正则化, 差分图像

Abstract: A method for automatically estimating parameters is applied to the L1-norm-based image super-resolution reconstruction framework. The approach can get a stable result with model errors and can improve convergence. A novel explanation of L1-norm-based bilateral total variation(BTV) term is presented according to pixel differences statistics, combined with probability prior model. Superiority of BTV is validated using Kullback-Leibler distance. A theoretical analysis of the feasibility and the problems of super-resolution reconstruction are given using mixture prior distribution.

Key words: super-resolution, regularization, pixel difference image

中图分类号: 

• TP18