论文部分内容阅读
SURF(Speed Up Robust Features)算法是对尺度不变特征变换SIFT(Scale Invariant Feature Transform)算法的一种改进,应用到遥感图像匹配领域中可以大大提高匹配速度,但是匹配精度略有下降。为此,本文提出一种基于无下采样Contourlet变换NSCT(Nonsubsampled Contourlet Transform)和SURF的遥感图像匹配算法。首先使用NSCT分别分解参考图像和待匹配图像,得到各自对应的低频分量;然后把这两幅低频分量图像作为SURF算法的输入图像进行预匹配,降低高频噪声对匹配结果的影响;最后利用预匹配结果求解变换模型的参数,并采用随机抽样一致RANSAC(Random Sample Consensus)算法剔除误匹配点对,解决了SURF算法存在的错误匹配问题。实验结果表明,与SIFT算法、SURF算法相比,本文算法具有更高的匹配精度和更快的匹配速度,且抗旋转、噪声、亮度变化能力更强。
The SURF (Speed Up Robust Features) algorithm is an improvement on Scale Invariant Feature Transform (SIFT) algorithm, which can greatly improve the matching speed in the field of remote sensing image matching, but the matching accuracy decreases slightly. To this end, this paper proposes a remote sensing image matching algorithm based on Nonsubsampled Contourlet Transform (NSCT) and SURF. Firstly, the NSCT is used to decompose the reference image and the image to be matched respectively to get the corresponding low-frequency components. Then, the two low-frequency components are pre-matched as the input images of the SURF algorithm to reduce the effect of high-frequency noise on the matching result. Finally, Matching results are used to solve the parameters of the transformation model. The Random Sample Consensus (RANSAC) algorithm is used to eliminate the mismatched point pairs, which solves the problem of false matching in SURF algorithm. The experimental results show that compared with the SIFT and SURF algorithms, the proposed algorithm has higher matching accuracy and faster matching speed, and is more robust against rotation, noise and brightness changes.