基于小波域改进SURF的遥感图像配准算法

Remote Sensing Image Registration Algorithm Based on Improved SURF in Wavelet Domain

为了进一步加快遥感图像配准速度,同时使其配准精度有所提高,提出了一种基于小波域改进加速鲁棒特征(speeded up robust features,SURF)的遥感图像配准算法.首先采用小波变换将基准图像和待配准图像分别分解获得其低频和高频分量;然后对低频分量提出改进SURF以得到粗配准点对:采用主成分分析(principal component analysis,PCA)对描述子降维,依据双向配准准则实现特征点的粗配准;接着利用两次距离阈值不同的随机抽样一致(random sample consensus,RANSAC)算法分级筛选出精配准点对;最后运用最小二乘法拟合几何变换参数完成配准.实验结果表明,与尺度不变特征变换(scale invariant feature transform,SIFT)算法、SURF算法、多尺度配准小波域SURF算法、基于NSCT(non-subsampled contourlet transform)和SURF的算法相比,本文算法不仅配准速度大大加快,同时配准精度也得到提高.

To further increase the speed of remote sensing image registration on the premise of improving registration precision,a remote sensing image registration algorithm based on improved speeded up robust features（SURF）in wavelet domain was proposed.Firstly,the reference image and the image to be registered were decomposed into the low-frequency component and high-frequency components by the wavelet transform,respectively.Then the low-frequency component served as the input image of improved SURF algorithm was proposed in this paper,thus obtaining the coarse registration point pairs.The dimension of descriptors was reduced by the principal component analysis method,and the coarse registration of feature points was achieved by the criterion of two-way registration.Next,precise registration point pairs were gradually screened out twice by using the random sample consensus（RANSAC）algorithm with different distance thresholds.Finally the parameters for geometric transformation were fitted through the least square method,and the registration image was obtained.A large number of experimental results show that the proposed method greatly improves the registration speed with higher registration accuracy compared with four algorithms,namely the scale invariant feature transform（SIFT）algorithm,the SURF algorithm,the multi-scale registration algorithm in wavelet domain using SURF,and the algorithm based on non-subsampled contourlet transform（NSCT）and SURF.