基于SIFT特征的图像配准与拼接技术研究

发布时间：2018-12-26 13:00

【摘要】：图像拼接与融合技术是图像处理和计算机视觉领域中的热门研究点,近年来已经被普遍地应用于国防安全、机器人视觉、视频监控、全景图像生成、医学图像处理、遥感、海底勘探、视频压缩与检索、3D虚拟场景构建等多个重要领域。其中图像拼接的配准率和拼接耗时以及图像融合的质量都是评价一个拼接方法好坏的指标。基于特征的图像拼接与融合方法因其配准的质量高并且不易受到尺度变化等因素的影响,受到广大研究人员的推崇。本文在现有的基于特征的图像配准与拼接技术的基础上,对图像配准和拼接融合的每一个步骤进行了细化和改进,最终得到了良好的配准与拼接结果。主要的工作如下:(1)分析比较了各种镜头畸变矫正算法的优缺点,根据实际需求采用张正友标定法或基于Ilya Krylov模型对摄像机参数进行了标定和矫正。(2)研究了各种图像增强的预处理方法,并提出了一种基于图像金字塔的增强算法,使得图像的配准率得到提升,在光照不佳的条件下配准率平均提高了14.3%。(3)研究了基于SIFT算法的图像配准,并改进了SIFT特征的描述符,使得图像配准的平均速度提高了2倍。(4)分析比较了各种图像融合方法,在原有的加权平均法上进行改进,使得融合结果的视觉效果有所改善。实验结果表明,文中提出的方法可有效用于各种领域的图像拼接和视频拼接。
[Abstract]:Image mosaic and fusion technology is a hot research point in the field of image processing and computer vision. In recent years, it has been widely used in national defense security, robot vision, video surveillance, panoramic image generation, medical image processing, remote sensing. Underwater exploration, video compression and retrieval, 3D virtual scene construction and other important areas. Among them, the registration rate, the time consuming and the quality of image fusion are all indicators to evaluate the quality of a stitching method. The feature-based image stitching and fusion method is highly praised by researchers because of its high registration quality and is not easy to be affected by the change of scale and other factors. In this paper, based on the existing feature-based image registration and stitching techniques, each step of image registration and stitching fusion is refined and improved, and good registration and stitching results are obtained. The main work is as follows: (1) the advantages and disadvantages of various lens distortion correction algorithms are analyzed and compared. According to the actual demand, the camera parameters are calibrated and corrected by using Zhang Zhengyou calibration method or Ilya Krylov model. (2) various image enhancement preprocessing methods are studied, and an image pyramid enhancement algorithm is proposed. The image registration rate is improved, and the average registration rate increases by 14.3% under the condition of poor illumination. (3) the image registration based on SIFT algorithm is studied, and the descriptor of SIFT feature is improved. The average speed of image registration is increased by 2 times. (4) the image fusion methods are analyzed and compared, and the original weighted average method is improved to improve the visual effect of the fusion results. Experimental results show that the proposed method can be effectively used in image mosaic and video mosaic in various fields.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】