摘要
对于从单站获得的目标图像进行轨迹及姿态的精确判读的问题,提出了一种新方法,即通过对图像中分割出来的目标图像,计算其惯量椭圆,并根据该惯量椭圆推算出目标的空间姿态角.这种方法充分利用了单站获得的二维目标图像的几何信息,并对目标图形的细节程度要求不高,适用于各种轴对称目标.在用vc ++实现了该方法后对几组试验图片进行了判读,并对实验结果做了对比分析后表明:方法准确度较高,结果稳定.
A new approach focused on how to measure the precision 3D pose of an object from mono-view had been put forward, which was based on calculating the ellipse of inertia of the object＇s projection that was obtained from an image through the image segmentation process and reckoning pitch and yaw angles of the object through this ellipse. It could be applied in any image containing objects of axial symmetry not only because it used the knowledge of object＇s figuration adequately, but also it did not require a high degree of image details which many other arithmetic dose. According to a conclusion that is drawn from analyzing experiment data which came from two series of images by using this approach, it is precise and stable.
出处
《光子学报》
EI
CAS
CSCD
北大核心
2007年第3期568-573,共6页
Acta Photonica Sinica
