多源遥感图像舰船目标SIFT匹配的性能分析

Performance Analysis of SIFT Matching in Ship Target Based on Multi-source Remote Sensing Images

基于多源遥感图像开展舰船目标识别是未来的发展趋势,但是SIFT特征匹配在该领域的应用性能尚不明确。利用双三次差值算法,模拟了6种空间分辨率等级的遥感图像,对舰船目标SIFT特征匹配进行了性能测试。结果显示:随着空间分辨率的下降,目标匹配对数会急剧减少,不同空间分辨率的舰船目标进行匹配时,可通过插值方法在一定程度上提高匹配性能。以上分析结果为多源遥感数据在舰船目标识别中的应用奠定了基础。

It is a future direction to identify ships based on multi-source remote sensing images,but SIFT （ Scale Invariant Feature Transform） matching performance is not clear in this research field.The ship targets of six spatial resolutions are simulated using bi-cu- bic interpolation method, and the SIFT feature matching of ship target are tested.The results show that the target matching number is sig- nificantly reduced as the spatial resolution decreases, and the matching performance can be promoted by interpolating when ship targets with different spatial resolutions are matched.The analysis results lay the foundations for application of multi-source remote sensing in ship target identification.