Design Of Subway Track Obstacle Detection System Based On Machine Vision

At present,rail transit plays an important role in the urban public transportation system.With the rapid development of urban rail transit,the subway is a proprietary,high-density,highvolume urban rail transit system that covers various underground and above-ground road rights in urban areas..my country has now opened a metro mileage of 6000 kilometers.As an important means of transportation for urban population,it greatly eases the pressure of urban ground transportation.Therefore,ensuring the safety of subway trains not only helps to maintain the normal operation of the subway,but also plays an important role in protecting the lives and property of passengers.The obstacles that may exist in the subway track environment are one of the important factors that affect the safety of subway trains.The subway driving environment may make it difficult for train drivers to respond effectively in time,which undoubtedly poses a threat to the safety of subway trains.Therefore,real-time detection of obstacles in subway tracks can ensure the safety of subway trains to a certain extent.Based on the current research on obstacle detection technology based on machine vision at home and abroad,according to the actual characteristics of subway driving environment and the actual needs of obstacle detection,this paper studies obstacle detection algorithms that are suitable for the experimental environment of this article,and builds the detection System architecture and system functional modules.The system uses the vehicle-mounted monocular machine vision method,through the analysis of the acquired video frame images,first identifies the track area,and then detects the obstacles in the target area.Specifically include the following:(1)Boundary recognition of subway track area.First,perform image preprocessing on the acquired video frame images to reduce the noise interference of the original image data;secondly,perform edge detection and orbit model construction on the processed images to improve the processing efficiency and detection accuracy of the algorithm.In the above processing,different algorithms are analyzed,and the optimal algorithm in this experimental environment is selected to identify the boundary of the track area.(2)Obstacle detection algorithm for subway tracks.This paper improves the existing interframe difference algorithm from the aspects of algorithm real-time performance and detection accuracy to complete obstacle detection.At the same time,the actual application of the convolutional neural network in this obstacle detection task is carried out,and the detection performance of the two is compared.(3)Design of obstacle detection system.It mainly includes the hardware system composition and software system design of the detection system.The hardware system includes video image acquisition equipment,video image processing equipment and video image display equipment.At the same time,the detection software architecture,system flow and functional modules are designed in detail.Through the above research,the system has been verified and analyzed in an experimental environment.The results show that the detection algorithm and detection system in this paper can detect obstacles on the track better,and have certain real-time and accuracy.