一种快速多模态图像匹配导航方案

A High Time-Efficiency Multi-modal Images Matching Scheme for Navigation

为了能在飞行器自主导航系统中快速,精确地完成不同模态图像间的匹配,该文提出了一种基于多分辨率分析的图像匹配导航方案。这种方案采用金字塔结构,由粗到细进行匹配。在最小尺度上,采用了逐步剔除的策略,提出了边缘膨胀模板(EDT),并推导了快速算法;在中间级别上,用Hausdorff distance作为相似准则;在最后一级别上,采用交互方差进行匹配。在匹配过程中,除最小尺度上采用快速算法外,在其他匹配阶段也推导和采用了一些加速算法来减少运算量。实验结果显示该文的方法可以快速,精确地完成多模态图像匹配。另外,该文方法易于硬件实现,有助于图像匹配导航系统的工程实现。

In this paper, a new scheme based on multi-scale is presented to match multi-modal images, in the airborne autonomous navigation system. Pyramid structure is exploited, in the scheme from coarse to fine. In the coarsest scale, a stepwise elimination scheme is introduced, the Edge Dilatation Template （EDT） is presented and some methods are introduced to reduce the calculation. In the middle scales, Hausdorff Distance （HD） is exploited as similar measure. In the finest scale, mutual variance is exploited. Some methods are also introduced and deduced to speed up the calculation in other matching scales besides the finest scale. The results of the experiments indicate that the multi-scale method can achieve fast, precision matching between multi-modal images. It can be realized by hardware easily and helps to the engineering realization of image matching navigation system.