The invention relates to a traffic light and road association method based on Bayesian classification, and the method comprises the steps: obtaining a traffic light and road association training set through training, enabling the input of the training set to be the association attribute of a traffic light and a road, and enabling the output to be whether the traffic light and the road are associated or not; inputting the association attributes of the to-be-tested traffic light and each road into the training set, and obtaining a preliminary association result of the to-be-tested traffic lightand each road according to a Gaussian naive Bayesian classification method; when the preliminary association result of the to-be-tested traffic light and the at least two roads is associated, the to-be-tested traffic light is detected; selecting a road corresponding to the final association result of the to-be-tested traffic light according to the preliminary association result; and using existingreliable truth value data of typical intersections as a training set. According to the method, a Bayesian classification method is used for obtaining the probability of relevance of all roads entering the intersection and all traffic lights in the intersection, and an incidence relation is established between the traffic lights and the road with the maximum relevance probability, so that the difficulty of artificially searching rules is avoided, and a better effect can be achieved.
本发明涉及一种基于贝叶斯分类的交通灯和道路关联方法,包括:训练得到交通灯和道路关联的训练集,训练集的输入为交通灯与道路的关联属性,输出为交通灯和道路是否有关联;将待测交通灯和各个道路的关联属性输入训练集,根据高斯朴素贝叶斯分类方法得到待测交通灯与各个道路的初步关联结果;待测交通灯与至少两个道路的初步关联结果是有关联时,根据初步关联结果选择待测交通灯的最终关联结果对应的道路,用已有的典型路口的可靠真值数据作为训练集,用贝叶斯分类的方法,得到进入路口的所有道路和路口内所有交通灯的关联性的概率,以交通灯和关联性概率最大的道路建立关联关系,这样既规避了人为寻找规律困难的难点,也能取得较好的效果。
Traffic light and road association method based on Bayesian classification
一种基于贝叶斯分类的交通灯和道路关联方法
2020-11-20
Patent
Electronic Resource
Chinese
TrafficNN: CNN-Based Road Traffic Conditions Classification
Springer Verlag | 2021
|Expressway road traffic state estimation method based on dynamic Bayesian network
European Patent Office | 2015
|Road traffic sign detection and classification
Tema Archive | 1997
|