在MAP超分辨率重建算法中, 相较于Gauss-MRF先验模型, Huber-MRF先验模型具有更好的保持图像边缘和细节的能力, 然而对Huber边缘惩罚函数的阈值选取一直没有更好的方式。在考虑红外图像细节纹理信息的基础上, 利用图像灰度共生矩阵, 把阈值参数与图像细节纹理信息联系起来, 实现对边缘惩罚函数阈值的自适应选取, 完成超分辨率图像重建。仿真实验证明, 该算法获取的高分辨率红外图像具有更高的信噪比, 而且更有效地保持了图像的高频信息。

In the MAP super-resolution reconstruction method, compared with Gauss-MRF prior model, the image edges and details can be preserved better in Huber-MRF prior model, but there is no better method for the threshold choice of Huber edge penalty function. In this paper, based on the infrared image detail texture information, the relation between the image detail texture information and threshold parameter is established by gray level co-occurrence matrix. The adaptive threshold choice of the edge penalty function is realized and the super-resolution reconstruction of image is implemented. The simulation experiment verifies that the infrared high-resolution images obtained by the proposed method have higher signal-to-noise (SNR) and preserve the high-frequency information of the images more efficiently.