改进模型估计与平差的多幅异源影像匹配方法

Volume Multi-Source Images Registration Based on Improvement Model Fitting and Adjustment

针对多幅异源影像匹配准确性不高的问题,提出了一种改进模型估计与平差的多幅异源影像匹配方法.首先利用基于快速自适应鲁棒性尺度不变的特征检测子与鲁棒性交叠的标准特征描述子,来增强特征匹配的鲁棒性;然后提出改进的随机抽样一致性算法,来提高模型估计的执行效率,同时保证鲁棒性;最后提出针对多幅异源影像匹配的平差方法,来优化异源匹配结果.实验结果表明,在异源影像存在较大差异的情况下,改进模型估计与平差的多幅异源影像匹配方法具有精度高的优势.

In order to improve the accuracy of volume multi-source images registration,a volume multi-source images registration was proposed based on improvement model fitting and adjustment(VMIRIMFA).Firstly,fast adaptive robust invariant scalable feature detector(FARISFD)and robust overlapped gauge feature descriptor(ROGFD)was used to enhance the robustness of feature registration.Then,improved random sample consensus was proposed to ensure the robustness of the algorithm and to improve the operation efficiency.Finally,an adjustment method was proposed for volume multi-source images registration and result optimization.The experiment results show significant advantages of VMIRIMFA in terms of precision for multi-source images with the large difference.