基于光流场分析的红外图像自动配准方法研究

提出了一种基于光流场分析的准确的红外图像自动配准方法．该方法可分为两个过程：先是利用全局光流场完成两幅图像背景区域的配准；其次利用由粗到细的层级匹配算法提取两幅图像中运动目标的特征点集，根据两组特征点集由最小二乘法计算出运动目标的变换参数，完成运动目标的配准．对一定研究领域的红外图像自动配准的仿真实验表明：该方法准确且对场景的运动有很好的鲁棒性．

RESEARCH OF AUTOMATED IMAGE REGISTRATION TECHNIQUE FOR INFRARED IMAGES BASED ON OPTICAL FLOW FIELD ANALYSIS

A novel accurate automated registration technique for infrared images based on the optical flow field analysis was presented. The technique can be preformed with two processes. Firstly the registration of background area of two images is finished based on the computation of global optical flow field. Then a hierarchical coarse-to-fine matching algorithm is used to extract the feature points of moving targets in two images. The registration of moving targets is finished using method of least squares which computes the transforming parameters of moving targets based on two set of feature points extracted. The simulation experiments for infrared images in some fields show that the method is accurate and robust to motion of scene.