【摘要】 在MAP超分辨率图像重建算法中,用Huber-Markov随机场(HMRF)作为图像的先验模型相比于Gaussian-Markov随机场(GMRF)能够更好地保护图像的边缘和细节。在以往的研究中,对于如何选取Huber函数的阈值参数T并没有一个很好的方法。本文提出了一种自适应的MAP超分辨率重建算法,该算法可以自动确定参数T,并根据重建的中间结果,不断对其进行更新,通过迭带最终得到重建图像。实验结果表明,该方法实现了参数的自动选取,在得到期望的高分辨率图像的同时,有效地保护了图像的边缘信息和细节。更多还原

【Abstract】 In the MAP super-resolution image reconstruction algorithm, the image discontinuities and details can be preserved better if Huber-Markov Random Fields (HMRF) is used as prior models compared with Gaussian prior models. There is no explicit method for the choice of Huber function parameter, or threshold T in the previous studies. In this paper, an adaptive MAP super-resolution reconstruction algorithm is proposed, in which the parameter T can be determined automatically and updated using the partially reconstructed result at each iteration step. The results of the experiments indicate that the proposed algorithm can not only make an automatic choice of the parameter T and get the high resolution reconstruction image expectedly, but also can preserve the edges and details of the image effectively.更多还原