【摘要】 基于对空中轴对称目标识别的需要,根据目标投影变换和姿态测量,利用匹配滤波和坐标变换,提出一种简化的识别方法。选择合适的目标投影图像作为识别模板,并用计算全息的方法编制成匹配滤波器;在目标的三维姿态被探测后,根据其方位角和俯仰角,对滤波器进行坐标变换,再利用光学相关方式识别目标的投影图像。通过数值模拟,分析了不同姿态下二维任意目标的识别情况,以及三维轴对称目标的模板选择及其识别效果。分析表明,利用这种简化识别方法,只需将目标在45°和90°俯仰角下的投影图像制成识别模板,并进行组合,即可得到目标的最小相关系数为0.4472,远大于飞机干扰物与模板的最大相关系数0.06。更多还原

【Abstract】 In order to recognize axes-symmetric targ et s in the sky, based on target′s projection transform and measurement of its thr ee dimensional (3D) orientation, a simplified approach is put forward in this pa per, which is realized in the way of matching filter and coordinate transform. A n appropriate projected image of a target is chosen to make a discriminating tem plate and is encoded to a matching filter by using computer-generated hologram. After 3D orientation of the target is explored, coordinate transform of the mat ched filter can be carried out according to its pitching angle and azimuth angle , and then this target is detected in the way of matching correlation. By this s implified detection method, it is studied how to select template for 3D axes-sy mmetric object, and recognized effect is simulated for 2D arbitrary target and 3 D axes-symmetric object at random angle. With the combination of only two filt ers, which are made from target′s projection of 45° and 90°, the minimum corr elation coefficient between the target and template in various pitching angle is 0.4472 which is far more than maximal correlation coefficient of 0.06 between p lane interference objects and template.更多还原