摘 要

在图像重建中，图像超分辨率重建占据关键位置。在图像采集和后期处理过程中，图像成像系统的硬件配置不完善，其他因素的影响使得图像重建效果达不到我们完全满意的程度。技术的发展和硬件成本的不断提高在提高硬件设备性能的基础上提高图像分辨率的方法已经不能满足人们的需求。因此，迫切需要从软件部分入手，研究提高图像分辨率的方法。

图像超分辨率重建技术是把低分辨率图像重建出高分辨率图像的技术。这个技术克服了硬件系统的局限性，易于实现，已经成为当前数字图像处理研究领域的热门之一。图像超分辨率重建的过程主要包括图像配准、图像插值和图像重建，其中，图像配准对重建效果影响最大。

本文重点研究图像配准算法。熟悉图像配准的原理和配准算法的分类，重点学习和研究两种典型的图像配准算法：SIFT算法（尺度不变特征转换算法）和SURF算法（加速稳健特征算法）。然后，通过仿真实验，对两种典型算法进行了分析比较，总结了典型算法的优缺点。最后，在实验分析的基础上，提出了一种基于SURF算法的改进算法。利用快速黑塞(Hessian)矩阵提取图像特征点,根据图像的熵信息对特征点进行滤波。接着通过随机抽样协议(RANSAC)消除误差配准。实验表明,改进后的算法能够有效地提高配准效率,提高配准精度。

关键词：图像配准；尺度不变特征变换；加速鲁棒特征；图像熵；

Research on registration algorithm of image super-resolution reconstruction

Abstract

In the process of acquisition and post-processing, the hardware configuration of the image imaging system is not perfect, and other factors play a key role in the image reconstruction. The influence of the image makes the image reconstruction effect less than we are completely satisfied. With the development of technology and the improvement of hardware cost, the method of improving image resolution has been applied on the basis of improving the performance of hardware equipment. Therefore, it is urgent to start from the software part and explore the methods that can improve the image resolution effectively.

Image super-resolution reconstruction is a technique for reconstructing a high-resolution image using multiple low-resolution images with complementary information in the same scene. This technology overcomes the limitations of the hardware system and is easy to implement. It has become one of the hot topics in the field of digital image processing research. The process of image super-resolution reconstruction mainly includes image registration, image interpolation and image reconstruction. Among them, image registration has the greatest impact on the reconstruction effect.

This article focuses on the image registration algorithm. First, the principle of image registration and the classification of registration algorithms are introduced. Two typical image registration algorithms are mainly studied: SIFT algorithm and SURF algorithm. Then, through simulation experiments, two typical algorithms are analyzed and compared, and the advantages and disadvantages of typical algorithms are summarized. Finally, based on the experimental analysis, an improved algorithm based on SURF algorithm is proposed. Using the fast Hessian matrix to extract image feature points, the feature points are filtered according to the image entropy information, and an improved fast nearest neighbor search algorithm is used to perform feature matching. Random sampling agreement (RANSAC) algorithm is used to eliminate false matching pairs. Experiments show that the improved algorithm can effectively improve the matching efficiency, and the matching accuracy can be improved.

Key words: image registration;Scale-invariant feature transform; Speeded-Up Robust Features;Image entropy；

目 录

摘 要 I

Abstract II

1 绪论 1

1.1研究背景及意义................................................................................................................. 1

1.2国内外研究现状 .................................................................................................................1

1.2.1国内研究现状...........................................................................................................1

1.2.2国外研究现状...........................................................................................................2

1.3本文研究的主要内容及组织结构......................................................................................2

1.3.1本文研究的主要内容...............................................................................................3

1.3.2技术路线...................................................................................................................3

1.3.3本文的组织结构.......................................................................................................3

2 图像超分辨率重建技术..............................................................................................................5

2.1图像超分辨率重建的基本原理 5

2.1.1图像超分辨率重建的概念...................................................................................... 5

2.1.2图像超分辨率重建的原理...................................................................................... 5

2.2图像超分辨率重建过程 5