【作者】 张妍；

【导师】 韩建峰；

【作者基本信息】 内蒙古工业大学 ， 信息与通信工程， 2020， 硕士

【摘要】 图像拼接技术是解决无人机航拍过程中成像范围与清晰度之间限制的关键技术,图像拼接算法的效率影响拼接图像的质量。但目前航拍图像拼接算法存在耗时长,准确率低等问题,拼接算法效率整体较低,对航拍图像拼接技术的应用产生了一定影响。本文围绕航拍图像拼接算法的改进与在公路监测中的应用两个内容进行研究。在图像拼接算法过程中,为了避免重叠区域外特征点对图像配准的时间以及匹配正确率造成影响,本文在特征点提取前计算图像间的有效区域。在有效区域内进行特征点的提取,避免无用特征点对配准结果产生的影响。为了提高图像间特征点匹配的准确率,本文利用稀疏表示中稀疏系数的唯一性进行特征点的匹配,避免图像间相似区域特征点的误匹配。在图像融合部分,结合空间域与变换域两种融合算法优势,改进现有的图像融合算法,对配准后的航拍图像进行融合,得到拼接图像,并通过实验证明拼接算法效率的提高。将航拍图像拼接算法应用于航拍公路路面监测图像的拼接中,解决目前监测手段具有局限性问题,提高监测效率。针对与现有车载系统图像处理上的不同,根据航拍路面特点,采用不同的图像处理方法,将拼接后的路面图像用于路面监测,实现基于无人机航拍图像拼接的公路路面监测。本文通过对航拍图像进行拼接实验,证明改进的拼接算法实现了降低图像拼接的时间,提高拼接准确率,改善拼接图像的质量。并将改进算法应用于公路监测的路面图像拼接过程中,改善现有监测手段的不足,为提高公路监测领域效率提供借鉴意义。更多还原

【Abstract】 Image stitching is a key technology to solve the limitation between imaging range and image resolution in the process of UAV aerial image.The efficiency of image stitching algorithm affects the quality of stitched images.However,the current aerial image stitching algorithm has low accuracy.The overall efficiency is low,which has a certain impact on the application of image stitching technology.This article focuses on the improvement of the aerial image stitching algorithm and its application in highway monitoring.In the process of image stitching algorithm,in order to avoid the influence of feature points outside the overlapped area on the time of image registration and matching accuracy,this paper calculates the effective area between images before feature points extraction.Through this,the mismatching of feature points in similar areas between images is avoided.In order to improve the accuracy between images,this paper uses the uniqueness of sparse coefficient in sparse representation to match feature points.This method can avoid the mismatching of feature points in similar areas between images and improve the registration accuracy.In the part of image fusion,combining the advantages of the two fusion algorithms in the spatial domain and the transform domain,the image fusion algorithm is improved.The aerial images are fused to obtain a stitched image,and the efficiency of the stitching algorithm is improved through experiments.The aerial image stitching algorithm is applied to the stitching of aerial road surface monitoring images to solve the limitation of current monitoring methods and improve the efficiency of monitoring.In view of the difference in image processing from the existing vehicle-mounted system,according to the characteristics of aerial image road surface,different image processing methods are adopted to use the spliced road surface image for road surface monitoring to realize highway road surface monitoring based on the aerial image stitching of UAV.Through the experiment on aerial images,this paper proves that the improved stitching algorithm can reduce the time,improve the accuracy of stitching,and improve the quality of stitched images.The improved algorithm is applied to the road image stitching process of highway monitoring,which improves the deficiencies of the existing monitoring methods and provides a reference for improving the efficiency of highway monitoring.更多还原