作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程

• •    

基于惯性椭圆拟合的小目标姿态测量方法

  • 发布日期:2020-03-06
  • Published:2020-03-06

摘要: 视觉测量方法在飞行器姿态测量中被广泛应用,然而远距离飞行目标成像纹理缺失会使传统光测手段应用受限。针 对这一问题提出一种小目标姿态测量方法,对测量中采用的追踪手段、图像分割技术、测量方法进行了研究。首先,利用高速 相机拍摄目标,并用跟踪算法对运动目标进行框选,然后使用非面积极大区域抑制的方法将目标分割出来,最后计算目标图 像的惯性椭圆实现目标姿态测量。针对只有 4 像素宽度的目标,利用仿真实验和真实场景模拟实验对测量精度进行了验证。 实验结果表明,在仿真情况下误差不超过 0.233 度;在真实拍摄低复杂度背景下误差不超过 0.307 度,在高复杂度背景下误 差不超过 0.325 度。基本满足导弹等小飞行目标姿态测量需求,测量结果稳定可靠,显著降低了判读工作量。

Abstract: The visual measurement method is widely used in aircraft attitude measurement. However, in terms of long-distance high-speed small target, the measurement error may be very large due to the lack of texture information. Thus, the application of the optical measurement method is limited. To address this problem, a small target attitude measurement method is proposed. In which the tracking methods, image preprocessing techniques and measurement methods used in the measurement were studied. Firstly, the high-speed camera is used to capture the target, then the target tracking algorithm is used to locate the moving targets, next the target with only the contour information is segmented by threshold segmentation and non-area maximal region suppression. Finally, the inertial ellipse of the target image is calculated to measure the attitude of the target. The measurement accuracy was verified by simulation experiments and real-world simulation experiments. The experimental results show that when the target is about 4 pixels wide in the image, the error is no more than 0.233 degrees in the simulation case; the error is no more than 0.307 degrees in the case of real low-resolution background, and the error is no more than 0.325 degrees in the high complexity background. The method satisfies the attitude measurement requirements of small targets such as missiles, the measurement results are stable and reliable, and the manual interpretation workload is significantly reduced.