基于图像处理和神经网络的站场状态信息识别方法

Recognition Method of Station Yard Status Information Based on Image Processing and Neural Network

为解决在部署位置追踪系统过程中遇到的站场状态信息来源问题,引入图像处理技术和神经网络进行站场状态信息识别。将站场显示图像作为数据源,先对图像进行降噪、位移纠正、分割等处理后,再提取图像的颜色特征作为神经网络的输入,实现对信号机、道岔和区段等设备不同状态的识别。经测试验证表明,本方法具有较好的识别效果,能够对外提供站场状态信息。

In order to solve the problem of status information source encountered in the deployment of train location system,image processing technology and neural network are introduced to recognize the status information of station yard.By using the station yard display image as data source,the image is processed by noise reduction,displacement correction and segmentation,and then the color features of the image are extracted as the inputs of the neural network to realize the recognition of different states of the signal,switch and section.The test results show that this method has good recognition effects and can provide status information for the external.