论文部分内容阅读
由于影像间显著的几何和辐射差异,多源遥感影像自动匹配一直是目前研究的难点问题。首先引入具有光照和对比度不变性的相位一致性模型,并对其进行扩展,构建相位一致性的特征方向信息,然后借助于梯度方向直方图的模板结构,利用其特征值和特征方向,建立一种局部特征描述符——局部相位一致性方向直方图(local histogram of orientated phase congruency,LHOPC),最后利用欧氏距离作为匹配测度进行同点名识别。对四组多源遥感影像进行试验,其结果表明,相比于尺度不变特征转换和加速鲁棒性特征算法,LHOPC能更为有效的抵抗影像间的辐射差异,提高了匹配性能。
Due to the significant differences in geometry and radiation between images, automatic matching of multi-source remote sensing images has always been a difficult problem in current research. First, a phase consistency model with illumination and contrast invariance is introduced and extended to construct the phase-consistent feature orientation information. With the help of the template structure of the gradient direction histogram and its eigenvalues and feature orientation, a The local histogram of orientated phase congruency (LHOPC) is used as the local feature descriptor. Finally, the Euclidean distance is used as the matching measure to identify the same point name. Four groups of multi-source remote sensing images are tested. The results show that LHOPC can effectively resist the radiation difference between images and improve the matching performance compared with the scale invariant feature transform and the acceleration robustness feature algorithm.