遥感影像超分辨率重建的字典学习类算法

发布时间：2018-03-21 05:51

本文选题：超分辨率重建　切入点：稀疏表示　出处：《测绘通报》2017年06期 　论文类型：期刊论文

【摘要】：近年来基于字典学习的超分辨率重建技术已成为图像处理领域的研究热点,相比基于重建的超分辨率方法,基于学习的方法充分利用了先验知识,在放大倍数较高时,仍可取得较好的效果,因此被公认为一种非常有前途的方法。本文对国内外已有的基于字典学习的超分辨率重建方法进行了系统研究,梳理了3种基于字典学习超分重建算法的基本原理及优缺点。此外,本文根据遥感影像的特点,使用同一数据源进行字典学习,利用不同字典学习算法分别生成高、低联合字典对,采用不同尺寸大小及缩放倍数的测试图像,进行超分辨率重建,对各种算法的重建性能、鲁棒性和复杂度进行综合分析,进一步研究了各种算法对遥感影像不同应用需求的适用性。
[Abstract]:In recent years, the super-resolution reconstruction technology based on dictionary learning has become a research hotspot in the field of image processing. Compared with the super-resolution method based on reconstruction, the learning-based method makes full use of prior knowledge, and when the magnification is high, Therefore, it is recognized as a promising method. In this paper, the existing super-resolution reconstruction methods based on dictionary learning have been systematically studied at home and abroad. In addition, according to the characteristics of remote sensing image, this paper uses the same data source for dictionary learning, and makes use of different dictionary learning algorithms to generate high. Low joint dictionary pairs are used for super-resolution reconstruction using test images of different sizes and scales. The reconstruction performance, robustness and complexity of various algorithms are analyzed synthetically. The applicability of various algorithms to different applications of remote sensing images is further studied.
【作者单位】：长安大学地质工程与测绘学院;地理国情监测国家测绘地理信息局工程技术研究中心;
【基金】：国家自然科学基金(41372330);国家自然科学基金青年基金(41601345)
【分类号】：TP751

【相似文献】