论文部分内容阅读
虽然合成来自多个传感器的数据以获得优良跟踪精度的方法是人们当前感兴趣的课题,而动态级融合的研究似乎仍然是一个悬而未决的问题。本文通过生成与公用状态矢量的量测相关的公用状态观察模型研究动态地来自雷达和成象传感器的数据融合。本文考虑的重点是雷达提供的反射中心量测值可能与真实目标中心不太相符这一事实。而且当目标方位改变时,反射中心可能会围绕真实目标中心变化。对远距离的小目标来说,这一影响可以忽略不计,但对半延伸或延伸(semi-extended or extended)的对象来说,这种失配就要加以考虑。本文研究的方法是根据球面坐标中真实目标中心未知恒定偏差的随机扰动,模拟雷达反射中心量测。这些偏差估计用于采用雷达和成象传感器数据跟踪目标的广义卡尔曼滤波方法中。产生跟踪滤波的性能由计算机仿真来评估。
Although methods of synthesizing data from multiple sensors for superior tracking accuracy are of current interest, the study of dynamic-level convergence seems to remain a pending issue. This paper studies the data fusion dynamically from radar and imaging sensors by generating a common state observation model associated with the measurement of common state vectors. The point to be considered in this paper is the fact that the reflection center measurements provided by the radar may not correspond well with the true target center. And when the target orientation changes, the center of reflections may vary around the true target center. This effect is negligible for small targets at long distances, but for semi-extended or extended objects this mismatch is to be taken into account. The method studied in this paper is to simulate the radar reflection center measurement according to the random perturbation of the unknown constant deviation of the true target center in spherical coordinates. These bias estimates are used in a generalized Kalman filter approach that uses radar and imaging sensor data to track targets. The performance of the generated tracking filter is evaluated by computer simulation.