为了提高多光谱图像匹配的速度和精度, 提出一种改进的随机抽样一致性(RANSAC)算法。针对传统RANSAC算法迭代次数多、运行效率低、单应性矩阵模型精度低等问题, 在采用SIFT 算法完成初始特征匹配的基础上, 从合理减少样本集中元素个数以提高局内点在样本中所占的比例以及采用预检验快速舍弃不合理的初始参数模型等方面对RANSAC算法进行改进, 从而极大地减少了算法的迭代次数, 提高了算法的运行效率和估计精度。实验结果表明, 所提改进算法不仅提高了图像匹配的精度, 而且在处理相同数据的前提下, 其所用时间不足传统RANSAC算法的60%, 有效减少了算法的运行时间, 提高了算法效率。

In order to improve the speed and accuracy of multi-spectral image matching, an improved random sampling consistency (RANSAC) algorithm is proposed. For the traditional RANSAC algorithm, it presents such problems as lots of iterations, low operation efficiency and low precision of the homography matrices estimation. In this paper, based on the SIFT algorithm to complete the initial feature matching, the RANSAC algorithm is improved by reasonably reducing the number of elements in a sample set, which can increase the proportion of intra-office points in the sample, and rapidly rejecting unreasonable initial parameter models by using pre-test. This can greatly reduce the number of iterations of the algorithm and improve the operation efficiency and accuracy of the algorithm. Experimental results show that the proposed method not only improves the accuracy of image matching, but also on the premise of the same data processing, it decreases the process time to be less than 60% of that of the RANSAC algorithm, which improves the efficiency of the algorithm.