提出基于离散动态贝叶斯网络模型,对若干可观测的目标特征参数进行综合推理。推导了离散动态贝叶斯网络的推理算法。建立了目标识别的离散动态贝叶斯网络模型。应用图形模式,使得计算量大大简化,降低了实用的复杂性。仿真结果表明,该方法能够将各种目标特征进行综合,使得各种特征及不同时刻的同一特征互相修正补充,克服了依靠单一特征进行目标识别的局限。
At the very beginning, a new inferring method was proposed, which synthesized different observable parameters of target characters based on discrete dynamic Bayesian network model. After that, a new inferring algorithm of discrete dynamic Bayesian network was discussed. In succession, the discrete dynamic Bayesian network model based on target identification was established. Then, graphic model to mitigate the complexity of the application was applied. Finally, the simulation result shows that this method can synthesize different target characters, enable them amend each other with respect to different time-phrases and accordingly overcome the limitations of identifying target merely through a single pattern.