用于遥感图像拼接的改进SURF算法

发布时间：2018-03-06 18:14

本文选题：计算机应用　切入点：图像配准　出处：《吉林大学学报(工学版)》2017年05期 　论文类型：期刊论文

【摘要】：经典的SURF算法存在许多不足,如特征描述符维度高、运算量大,对于旋转和拍射视角变换角度过大时,匹配精度低等。针对以上问题,提出了一种改进算法,首先通过Hessian矩阵提取特征点,然后采用特征点圆形邻域进行特征描述,使用Haar小波响应为每个特征点建立描述符,同时计算邻域内归一化的灰度差分及二阶梯度,形成新的特征描述符,最后采用RANSAC算法剔除误匹配点。该算法不仅较经典SURF算法具有速度优势,同时充分利用了灰度信息和细节信息,具有更高的精度。实验结果表明:该算法对图像的模糊、光照差异、角度旋转、视场变换等均有良好的鲁棒性和稳定性。将该算法应用于遥感图像拼接,得到无明显几何移位、边缘衔接良好的拼接图像。该算法是一种耗时短、精度高的图像配准算法,能够满足遥感图像拼接对配准的要求。
[Abstract]:The classical SURF algorithm has many shortcomings, such as high dimension of feature descriptor, large amount of computation, low accuracy of matching when the angle of rotation and angle of view is too large, etc. In view of the above problems, an improved algorithm is proposed. Firstly, the feature points are extracted from the Hessian matrix, then the feature description is carried out by using the circular neighborhood of the feature points. The descriptor for each feature point is established by using the Haar wavelet response. At the same time, the normalized grayscale difference and the second-order gradient in the neighborhood are calculated. A new feature descriptor is formed, and the RANSAC algorithm is used to eliminate the mismatch points. This algorithm not only has the advantage of speed over the classical SURF algorithm, but also makes full use of the gray level information and detail information. The experimental results show that the algorithm has good robustness and stability for image blur, illumination difference, angle rotation, field of view transformation, etc. The algorithm is applied to remote sensing image mosaic, and no obvious geometric shift is obtained. The algorithm is a kind of image registration algorithm with short time consuming and high precision, which can meet the registration requirements of remote sensing image stitching.
【作者单位】：中国科学院长春光学精密机械与物理研究所;中国科学院大学;
【基金】：吉林省重大科技攻关项目(11ZDGG001) 装备预研项目国家林业公益性行业科研专项项目(201204515)
【分类号】：TP751

【相似文献】