基于决策树的立交桥结构识别

Overpass structure recognition based on decision tree analysis

立交桥是道路网重要的组成部分,正确识别立交桥结构有助于道路匹配、导航规划和交通能力评估等方面旳研究。针对模板库识别立交桥方法的不足,本文提出一种基于决策树和构成规则的识别方法,利用智能算法学习匝道特征,进而识别和描述立交桥结构。该方法首先构建立交桥范圈圆逸取相关路段,利用路段形态特征和连接关系区分主干路段与匝道路段,然后使用决策树对匝道进行分类,最后依据不同类型立交桥的构成规则实现立交桥结构识别。试验结果表明,该方法可以有效识别和描述立交桥。

Overpass is an import a pari of the road network.Correct recagiiitipn of overpass stniflures contributes to road matching,navigation and traffic capability assessment.Aiming at the sliorlcomings of the template library in recognizing overpass structures,a recognition nielhod based on decision tree analysis and construction rilles is proposed which adopts intelligent algorithms to learn the characlerislics of the ramps,a?d recognize and describe the overpass structures.Firstly,the relative roads are selected by conslructiiig an overpass range circle and the road segment morphological features and connection relationsliips are used to distinguish the main roads and ramps.Secondly,the decision tree is used to classify the ramps and finally the constnic-tion rules of different types of overpasses are used to recognize the overpass structure.The test results show that the method is valid to recognize and describe the overpass stnicture.