应用图像形状分析的飞机目标方位角识别算法

Determination of aircraft aspect angle based on shape analysis

飞机姿态信息与其速度、加速度之间的密切联系决定了利用该信息可以更精准、更及时地估计目标的运动学属性。飞机姿态信息可以通过分析成像导引头获取的图像得到。针对红外图像中飞机目标的机头方位角识别问题,提出了一种适用于任意成像姿态角条件下的飞机方位角识别策略。首先求取飞机轮廓的距离-角度标记图,进行中值滤波和角度拓展预处理,再利用最小二乘曲线拟合方法检测标记图曲线的峰值点。基于飞机机头、机翼、机尾空间分布的先验知识,结合标记图峰值点幅值和角度的逻辑关系进行推理,识别出飞机方位角。仿真实验分析表明:与传统Hough变换方法相比,该算法具有更广泛的适用性,识别概率可达到90%以上,且简单高效。

Pose information of the aircraft target embodies its kinematic characteristics.The pose can be estimated from mono-ocular image data,because of the close relationship between the imaging shape and the relative pose.Aiming at the problem of aspect angle identification,a new valid algorithm which was suitable for arbitrary viewing direction was presented in this paper.Distance-angle function signature of the aircraft silhouette was figured out,then preprocessing including median filtering and angle expanding was performed.At last the peak of the signature curve was detected by using the least squares curve fitting method.Based on the distribution characteristics of the head,wing and tail of the aircraft,the aspect angle of the target could be recognized via the logical relationship between the peaks′ amplitude and angle.Experimental results show that the proposed algorithm is simple and has a more widely applicability than the traditional methods,and the identification probability can reach above 90%.