马尔可夫模型与Shearlet变换结合的SAR图像超分辨率复原方法

SAR Image Hallucination Based on Markov Model and Shearlet Transform

为了不改变成像硬件条件,通过软件方法提高SAR图像分辨率,提出一种马尔可夫随机场(MRF)模型和Shearlet变换相结合的超分辨率复原方法。该方法分为两个过程,训练过程和学习过程。在训练过程中,首先对训练库中的高、低分辨率图像进行Shearlet变换,提取不同方向、不同分辨率的中、高频信息,然后对不同方向的中、高频信息进行分块。在学习过程中,使用Shearlet变换提取待复原图像的中频信息并对其分块,然后在训练库的辅助下,使用MRF建立图像特征模型,最后通过最大后验概率(MAP)估计出各个方向的高频信息,将估计出的高频信息和待复原的低分辨率图像叠加到一起进行Shearlet反变换,最终获得高分辨率图像。通过对真实SAR图像的处理结果表明,无论是主观的视觉效果还是客观的指标上,本文提出的方法都取得较好的结果,优于传统插值方法以及目前最新的基于稀疏表示的超分辨率方法。

To enhance the resolution of SAR image,based on Markov model and Shearlet transform,a learning based super-resolution algorithm was proposed.The proposed method consisted of two stages of training stage and learning stage.In the training stage,firstly,Shearlet transform was performed to high-resolution and low-resolution images in the training set to obtain high-frequency and mid-frequency information of different directions.Then these high-frequency and mid-frequency information were divided into blocks.In the learning stage,Shearlet transform was performed to extract the mid-frequency information of a low-resolution image.Then,Markov network was adopted to model the super resolved high-resolution image with the blocks obtained in the training stage.Maximum A Posteriori（MAP）was used to estimate the high-frequency information of the low-resolution image in different directions.The estimated high-frequency information and the low-resolution image were transformed into super resolved high-resolution image through inverse Shearlet transformation.Experimental results on SAR images showed that the results of the algorithm have a good performance in terms of visual effects and root mean square error.