基于Gibbs分布的盲图像修复

针对盲图像去卷积问题，首先建立了一个基于Gibbs分布加权算法模型，然后讨论了关于能量泛函极小解的存在性，最后导出了相应的热流。在计算机模拟中，除了某些限制条件外，即使对于点扩散函数和原始的真实图像没有更多精确的信息，此算法也运行得很好。为了进一步观察此算法的实效性，与全变差盲去卷积模型的实验结果进行了比较。

Blind image deconvolution based on Gibbs distribution

A weighted total variation algorithm was proposed for solving some blind deconvolution problems, and then we got the existence of energy functional minimum and educed corresponding thermal flow. In computer simulations, the algorithm performs well even when we have no more accurate prior knowledge about point spread function and the original true image, besides some other constrains. Experiments show that the proposed algorithm is more effective and performs better than the total variation blind deconvolution method proposed.