Super-resolution reconstruction method based on adaptive-regularization
-
摘要: 传统正则化超分辨重建得到的图像往往存在过度平滑或伪信息残留的问题,结合超分辨重建模型对重建图像伪信息的产生进行了分析,针对传统方法的不足提出了基于图像区域信息自适应的正则化方法,通过图像的区域信息将图像划分为平滑区与非平滑区域,对不同区域选用不同的先验模型进行约束。同时考虑人眼的视觉感知特性,结合区域信息实现正则化参数的自适应选取。实验结果表明该方法在抑制重建图像伪信息的同时能有效保护细节,效果要优于传统方法与单一的先验模型约束,对于红外与可见光图像重建效果的提升提供了一定的理论参考。Abstract: The images reconstructed by traditional regularization super-resolution often have over smoothing or different artifacts residue. The cause of artifacts is analyzed by super-resolution reconstruction model. To improve the disadvantage of traditional methods, this paper proposes an adaptive regularization algorithm based on image region information, the original image is divided into smooth and non-smooth regions by the information, each type of region use different type of prior model as constraints. Considering the characteristics of human vision, regional information is used to achieve adaptive regularization parameter selection. Experiment results indicate that the proposed algorithm can improve the quality of reconstructed image with better artifacts smoothing and details preserving than traditional method and regularization with single prior model, which provides a theoretical reference to enhance infrared and visible light image super-resolution reconstruction quality.
点击查看大图
计量
- 文章访问数: 1181
- HTML全文浏览量: 270
- PDF下载量: 590
- 被引次数: 0