针对重复纹理区低空影像匹配中存在的多义性易造成同名点过少或匹配失败等问题,利用相位相关法计算了像对中两影像傅立叶变换复共轭的互功率谱,预测同名点在平移、旋转和尺度上的遍历范围,并以尺度遍历范围自适应估计Harris-Laplace尺度空间和检测角点,然后通过相关系数法和核线约束匹配同名点。实验验证了该方法对于重复纹理区低空影像匹配的可靠性和实用性,可获得数量足够且分布均匀的同名点。
Low altitude image matching with repeated texture may have a small amount of corresponding points or beunsuccessful because of ambiguous matching. To solve the problem, a phase correlation method supportes low altitude aerial image matching with repeated texture is proposed. The method estimates the traversal range of corresponding points, including translation, rotation, and scale space. The traversal range is estimated by a cross-power spectrum of the image Fourier trans- form and a complex conjugate of another image Fourier transform. Next, Harris-Laplace scale space is estimated by the scale traversal range and corners detection. Then, corresponding point matching is achieved by a correlation coefficient and epipolar constraint. The experimental results show that the method may be reliable and practical for low altitude images matching with a repeated texture, and obtains enough and well-distributed corresponding points.