用于图像重建的图像快速高精度匹配算法研究分析.docVIP

用于图像重建的图像快速高精度匹配算法研究 中文摘要 图像重建是计算机视觉研究领域的重要技术，它是从多幅具有相互位移且有互补信息的图像中获得一幅更高分辨率图像的过程。它通过对同一目标拍摄多幅图像，并利用图像配准算法求出由于摄像机抖动引起的每幅图像的位移量，最后利用重建算法提高图像的分辨率。由于该项技术能大幅改善图像质量，因此在视频监控、医学诊断、遥感、天文观测、生物信息识别等领域有很好的应用前景。 为了提升实际应用中图像重建速度，本文对快速配准算法进行了重点研究，研究内容主要包括特征提取、特征点定位、特征匹配算法及重建、基于OpenCV平台实现等问题。 本文主要工作如下： 1. 针对传统图像特征提取算法速度慢的问题，在分析对比各特征提取方法和策略的前提下，提出了一种快速的特征提取算法，该方法速度快、精度高，可以满足准实时图像重建的要求。 2. 深入研究了Harris角点定位的原理，将角点的定位精度提升到亚像素级，同时利用金字塔Lucas-Kanade算法进行匹配以解决传统算法处理大而不连贯的运动时匹配不准的问题，最后采用RANSAC算法去除伪角点对，提高匹配准确度。 3. 综合分析比较各重建算法性能后，采用维纳滤波去噪方法进行图像复原。 4. 基于OpenCV软件平台，完成图像特征点提取、特征匹配、 最后总结了本文所做工作，并对今后图像配准研究方向进行展望。 关键词：图像重建；亚像素；维纳滤波；OpenCV；特征匹配 Research On The Fast And High Precision Matching Algorithm For Image Super-resolution Reconstruction Abstract Image reconstruction is an important technology in the field of computer vision research, which is a process that puts multiple images with displacement and complementary information into a higher resolution single image. Firstly, the technology obtains multiple images for the same target; secondly, it uses the image registration algorithm for calculating the displacement of each image caused by camera shake; finally, a higher resolution single image can be obtained by the reconstruction algorithm. Because the technology can significantly improve the image quality, Image reconstruction has a good application prospect in many fields, such as video monitoring, medical diagnosis, remote sensing, astronomical observation, biological information identification and so on. In order to meet the speed of image reconstruction in practical application, the research work in this paper focuses on fast registration algorithm, including feature extraction, feature point positioning, feature matching algorithm and implementation based on OpenCV platform. In this paper, the main contribution can be summarized as follows: 1. According to the low speed of feature extraction using traditional algorithms. With an analysis of the feature extraction algorithms and strategies