研究了飞机在天空飞行图像中的飞机跟踪问题.在分析飞机在天空飞行图像特点的基础上,提出了联合光流技术和区域轮廓模型跟踪运动目标的方法.首先,对图像进行高通滤波预处理,提取一些高亮度的斑点,获得运动目标的特征图像;然后用特征图像求解运动目标的光流,以光流对运动区域进行分割,获得运动区域的中心和半径,以该中心和半径的圆作为水平集的初始曲线;再采用Songand Chan方法快速检测运动目标的边缘.实验证明,该方法能够快速、准确地自动跟踪运动目标.
The paper addresses the issue about how to track the airport in images of airport flying in the sky. By analyzing the characteristics of sequence images that airport flying in the sky, it is proposed that tracking approach moving target using optical flow and region-based active contour. Firstly, the image is processed through high filter and the feature image of moving target is obtained. Secondly, the optical flow of moving target can be solved by using feature image. Finally, Moving area is segmented by using optical flow and the center and radius of moving area are obtained. The circularity of the center and radius turns to the inital curve of level set. The Song and Chan approach is implemented to detect the edge of moving target rapidly. It is demonstrated in this experiment that moving target can be tracked automatically, rapidly and accurately using the method.