图像增强的自适应网格交互式多模型算法

自适应网格交互式多模型算法(AGIMM)单纯依靠目标的模型后验概率来调整模型集,因此造成了在目标机动时刻跟踪性能不高的问题。基于图像传感器对目标机动的快速检测性能,首次将图像中的目标姿态角信息引入到该算法中。通过建立AGIMM估计的模型角速度与姿态角蕴含的模型角速度之间的隶属度关系,修正了目标的模型后验概率,增强了模型辨识能力,提高了该算法在目标机动时刻的跟踪性能,从而实现了图像传感器与雷达传感器两种异类信息的有效融合。

Image Enhanced Adaptive Grid Interacting Multiple Model Algorithm

Since the Adaptive Grid Interacting Multiple Model Algorithm (AGIMM) adjusts the model-set depending only on the target posterior probability，the tracking performance is not well when the target is in maneuvering.Considering the quick maneuvering detection performance of the image sensor，we firstly introduced the attitude information of target obtained from image in the AGIMM.By establishing the membership relation between the angular velocity estimated from the AGIMM algorithm and the model angular velocity from target attitude，the posterior probability of the model was modified，the model recognition ability was enhanced，and the tracking performance when target is maneuvering was improved.The fusion of different information from image sensor and radar sensor was realized effectively.