基于分类模型的目标跟踪算法研究

基于分类模型的目标跟踪算法采用分类模型实现对多目标的跟踪。在传统算法中目标跟踪的对象是每个传感器的实时采样,而在分类模型中将传感器采样根据分类规则进行分类。然后将类的特征作为一个新的采样为下一层的模型提供输入,可以减少计算的复杂度。该模型是一个可扩展模型,可以为分类器设计不同的算法对模型进行扩展。模型自身采用低耦合的层次化设计,每一层均可以采用不同的分类器,这样可以充分利用各种分类算法的优点。

Research of Target Tracking Algorithm Based on the Classification Model

The algorithm of target tracking is based on the classification model.In this algorithm the target was not only the samples from the sensors.Samples were classified by rules which were defined by the model.In every class a special value was calculated which was considered as a new sample and it was the input of the next level.The classifiers can be designed with different algorithms and the model can be combined with different classifiers.