摘要
由不同传感器摄取的遥感影像因成像模式、拍摄角度和分辨率不同,给两者之间的配准造成困难。针对该问题,提出归一化SIFT算法,通过对SIFT描述子归一化的处理,降低不同光学影像色调差异大的影响,并通过与最小二乘法和双线性内插法的结合,完成自动配准。选取角度和尺度偏差较大的SPOT与ASTER影像、ASTER与TM影像2组数据进行实验。结果证明,该算法鲁棒性强,配准精度高。
The imaging model,acquired angle and resolution between image acquired by diverse remote sensing sensors are different,which throws difficulty in the registration between them.Aiming at this problem,the normalized SIFT algorithm is proposed in this paper.The SIFT descriptors are normalized,which can reduce the impact of hue difference between image acquired by different sensors,and then combined with least-square equation and bilinear interpolation method,the automatic registration is achieved.Two groups of images,which have big differences in acquired angle and resolution,SPOT and ASTER,ASTER and TM,are tested.Results dedicate that this algorithm is robust and has a high accuracy.
出处
《计算机工程》
CAS
CSCD
北大核心
2010年第19期21-23,共3页
Computer Engineering
基金
国家“863”计划基金资助项目“卫星遥感SAR与光学影像自动配准与融合技术系统研究”(2007AA12Z157)
中国科学院知识创新工程青年人才领域前沿项目专项基金资助项目(O8S01100CX)
国家自然科学基金资助项目“基于背景学习的并行粒子滤波红外弱小目标TBD算法研究”(40901234)
关键词
不同光学影像
自动配准
归一化SIFT描述子
different optical image
automatic registration
normalized SIFT descriptor
