图像特征识别的m序列列车定位方法

Train Positioning Method by m Sequence Based on Image Identification

提出一种图像特征识别的m序列列车定位方法,利用m序列(最大周期线性反馈移位寄存器序列)对轨道上的绝对位置进行编码,系通过安装在列车顶部的摄像机依次拍摄由二进制特征标志组成的轨道位置编码。提出基于虚拟轨迹获取目标图像方法,解决同一二进制特征标志重复采集的问题;提出基于BP(向后传播)神经网络的位置编码识别方法,以处理轨道现场环境干扰严重的问题。不同位置上的编码具有唯一性,以此位置编码值作为地址与车载数据库位置信息进行对比,来实现定位。试验结果表明,该方法可以有效地实现列车定位。

A train positioning method by m sequence based on image identification is put forward, it uses the m se- quence linear feedback shift register （maximum cycle se- quence） to encode the absolute position on tracks, which consists of the binary characteristic symbols taken by cam- eras installed on the top of train successively. A method to obtain target imagebased on virtual path is proposed to solve the repeated collection problems of the same binary feature symbols, then the location coding identification method based on BP （back propagation） neural network is proposedto deal with the issues of serious environmental in- terference in track field. Different positions of coding havethe uniqueness, this position encoding value is taken as the address compared with vehicle location information da- tabase to realize train positioning. The test result shows that the method can effectively realize the train positioning.