采用频闪成像技术的柔性气动外形运动物体SIFT特征跟踪(英文)

SIFT features tracking for moving object with flexible aerodynamic shape based on stroboscopic imaging technique

近年来,有关微型飞行器的结构动力学、材料气动弹性以及飞行控制方面的研究受到了高度的重视。为了提高飞行器的空气动力学检测能力,提出了一种基于频闪成像技术的柔性气动外形物体跟踪方法。利用尺度不变特征转换(SIFT)算法提取出的特征对于图像的尺度变换、旋转以及光照变化和局部图像变形等具有的不变特性,提出了一种利用感兴趣区域中SIFT特征对柔性气动外形运动物体进行探测与跟踪的方法。该方法在获取物体的状态、位置以及空间转换关系等方面表现出良好的性能,并且为同一场景中同一物体在不同位置之间的相互匹配提供了可靠保证。实验表明:在该实验系统中,基于频闪成像技术、利用SIFT特征作为柔性气动外形物体探测的方法具有一定的可行性。

The research on structural dynamics, material aeroelasticity and flight control of a micro aircraft has become an important subject in recent years. In order to improve the detecting method for aircraft aerodynamic characteristics, a tracking approach for objects with flexible aerodynamic shape based on stroboscopic imaging technique was proposed. As scale invariant feature transform （SIFT） algorithm was invariant to image scaling and rotation, illumination changes and local deformations, a method that used SIFT features of any region of interest was presented to detect and track a moving object. The results show that the method is powerful in obtaining object state, location and spatial interrelation. Feature points are adequate to perform reliable matching between different locations of an object in a scene, moreover, to achieve object recognition. Experiments demonstrate that it can achieve good performance using SIFT features based on flexible aerodynamic shape as a detection method in the system.