基于频闪成像技术的MAV气动外形三维视觉测试(英文)

Three-Dimensional Visual Measurement for MAV Aerodynamic Shape Based on Stroboscopic Imaging Technique

在目前的实验条件下虽已掌握了大量的气动外形测试数据和分析结果,但这些实验都是基于高速影像设备来完成物体运动记录的.为此,提出了一种基于频闪成像探测技术的柔性气动外形三维视觉检测方法,基于此方法,能够在不具备高速影像设备的条件下获取微小型飞行器(MAV)柔性气动外形的连续图像数据,使用尺度不变特征转换(SIFT)方法提取出描述表面纹理特性的特征点,在频闪图像中的各个不同位置依次进行搜索、匹配,找出特征点的同名点.为了提高跟踪精度,使用了RANSAC方法对误匹配对进行消降.实验结果表明,该方法不仅能实现柔性外形的形态学变化检测,而且能建立一幅频闪图像中不同位置上提取出的特征点之间的关系,从而计算出气动外形上特征点的运动轨迹.整个实验框架是基于双目立体视觉测量原理来满足检测与跟踪的高精度及可视性的.

Current researches have obtained massive experimental results and analyses, however, they mainly use the high-speed image equipment to record object movement. This paper introduced a new three-dimensional visual measuring technique for flexible aerodynamic shape, which is based on stroboscopic imaging detection technique. By using this technique, continuous image data of micro air vehicle （MAV） aerodynamic shape during its movement were obtained in the situation in which there was no high-speed image equipment. In the experiment, features that describe surface texture information were extracted by using scale invariant feature transform （SIFT）. Through matching, those points which are the same as the extracted SIFT features at different locations in stroboscopic image were searched. In order to improve the accuracy of tracki[ng, RANSAC algorithm was applied to eliminating incorrect pairs. As a result, the corresponding relatienship among extracted points at different locations in one picture was established and the morphological change value, then motion trace of feature points extracted from aerodynamic shape during movement could be calculated. The whole experiment framework was based on binocular vision measurement principle to obtain sufficient visibility and enough accuracy for tracking.