基于区域分块与尺度不变特征变换的图像拼接算法

发布时间：2018-01-18 01:12

本文关键词：基于区域分块与尺度不变特征变换的图像拼接算法　出处：《光学精密工程》2016年05期 　论文类型：期刊论文

【摘要】：针对图像匹配算法计算量大,实时性差的问题,提出了一种基于区域分块与尺度不变特征变换(SIFT)相结合的图像拼接算法。该算法利用图像能量的归一化互相关系数快速分割出匹配图像与待匹配图像间的相似区域,利用SIFT算法在重叠区域中搜索出能用于匹配的图像特征点并实现快速精确配准。然后,通过对图像进行了几何校正和图像融合来实现图像序列间的无缝拼接。实验结果表明,该算法减少了传统SIFT算法的大量无用搜索,改善了图像的几何失真,降低了算法复杂度,提高了图像匹配的速度,在保证90%以上的匹配准确率的基础上,计算时间较传统SIFT算法减少了近50%。提出的算法可准确、快速地实现有形变和尺度变换图像的无缝拼接。
[Abstract]:In order to solve the problem of large amount of computation and poor real-time performance of image matching algorithm. In this paper, a new method based on region partitioning and scale-invariant feature transformation (sift) is proposed. The algorithm uses the normalized correlation number of image energy to quickly segment the similar region between the matching image and the image to be matched. The SIFT algorithm is used to search the image feature points which can be used to match the overlapped region and to achieve fast and accurate registration. Through geometric correction and image fusion, the image sequences are jointed seamlessly. The experimental results show that the algorithm reduces a large number of useless searches of the traditional SIFT algorithm. The geometric distortion of the image is improved, the complexity of the algorithm is reduced, the speed of image matching is improved, and the matching accuracy is guaranteed over 90%. Compared with the traditional SIFT algorithm, the computational time is reduced by nearly 50 points. The proposed algorithm is accurate and fast to realize seamless stitching of images with deformation and scale transformation.
【作者单位】：沈阳航空航天大学电子信息工程学院;中国科学院长春光学精密机械与物理研究所;长沙航空职业技术学院;
【基金】：国家自然科学基金资助项目(No.61171081) 辽宁省自然科学基金资助项目(No.2013024008) 中航工业航空科学基金资助项目(No.20122654004)
【分类号】：TP391.41
【正文快照】： 1引言图像拼接是将相互有重叠区域的图像进行配准、融合,从而得到一幅大视野,高分辨率的新图像。图像拼接的主要步骤分为图像预处理、图像匹配和图像融合,核心过程为图像匹配,即确定两幅图像中的重叠位置,匹配的精确性将直接决定拼接成功与否。匹配精度高、速度快的匹配算法在

【相似文献】