Research On Method For Monitoring Railway Subgrade Settlement Based On Computer Vision

All along,railways play an important role in promoting economic development and connecting transportation hubs.However,due to the severe weather for a long time,the roadbed of the railway is prone to deformation and settlement.When the settlement has accumulated to a certain extent,it is easy to cause major accidents such as train derailment.Therefore,the railway road condition monitoring is particularly important,and the real-time monitoring of subgrade subsidence is required,and the results should be accurate.At present,the commonly used methods for monitoring settlement of subgrade are monitoring pile method,settlement plate method,layered settlement method and GPS measurement method.These methods have some defects.The monitoring pile method and the settlement plate method belong to manual measurement,which have large working intensity,repeated labor and low automation.They can not be measured during the driving process,and their measuring accuracy is easily affected by manpower.The law of layered settlement has a great dependence on the environment,it will hinder the progress of construction,and the measurement accuracy it can achieve is also easily affected by the external natural factors.The disadvantage of the GPS method is that the measurement accuracy is limited,and the most important is the high cost.Therefore,a monitoring method of railway subgrade subsidence based on computer vision is designed in this paper.Compared with other common methods,this method has the characteristics of automatic and high precision,and it can also run effectively during the train moving.The method is respectively provided with a fixed identification board on the roadbed of the railway and the base thereof,and the purpose of monitoring the settlement degree of the railway subgrade is achieved by measuring the distance change between the signs.The method is mainly divided into two modules: the identification point extraction and the distance calculation between the subgrade and the base marker.In order to extract the marked points on the identification board,the mark point is designed as a circle.The mark point extraction module,first with a camera installed signage of the railway subgrade photos,and then extract the pixel edge identification point by using the edge detection operator,then use the method to improve the sub-pixel edge positioning accuracy,finally using the method of least squares fitting algorithm to obtain the identification of point coordinates.In the calculation module of subgrade and base marker distance,this paper designed and implemented the distance measurement method based on binocular vision and monocular vision respectively.Finally,from the experiment and theory,two kinds of distance measurement methods are analyzed and compared,the distance measurement method based on binocular vision and anti noise ability,method of measurement error and error fluctuations is relatively small,but the difference of two methods to measure the effect is not significant,considering the factors such as the efficiency of processing system and,this paper chose distance measurement method based on monocular vision for railway subgrade settlement monitoring.