近似核线影像的轻量级多阶多尺度特征准稠密匹配

A MWVD-based Algorithm for Point Clu

  • 摘要: 针对近似核线影像,设计了轻量级的线状多尺度三角塔结构,通过多尺度下复合特征的极值检测,得到大量稳定特征点,进行特征点尺度和位置的拟合以提高算法对影像变形和核线几何计算误差的适应性,然后对特征点进行复合分量的综合描述,并采用属性特征和数值特征相结合的方法在预测区间进行快速匹配,最后采用微分纠正法提高同名点定位精度。实验表明,本算法能够快速得到分布均匀、密度较大的可靠同名点精度约0.1像素级。

     

    Abstract: A new algorithm based on a multiplicative weighted Voronoi diagram(MWVD) was proposed in this paper.The idea of the algorithm is to: select appropriate factors for describing the statistical,thematic,topological and metric information,and integrate the factors in the process of point feature generalization to ensure different types of information may be transmitted correctly,and the generalization of point clusters is done by repetitively constructing MWVDs.

     

/

返回文章
返回