Study On The Railway Tunnel Entrance Obstacle Detection Method Based On Structured Light Measurement Technology

The railway tunnel entrance is an important traffic area for the safe operation of trains,and the obstacle detection is an important means to ensure the safety of the railway tunnel entrance.In consideration of the limitations of the existing obstacle detection techniques and methods in the railway tunnel entrance,the railway tunnel entrance obstacle detection method using structured light measurement technology was proposed,focusing on how to effectively implement the obstacle detection image processing algorithm under several weather conditions.The study results of this thesis are as follows:(1)In view of the interference problem of the sunlight on the obstacle structure light image in the railway tunnel entrance scene,an image processing method adopting the background subtraction of the point structured light image was proposed.According to the characteristic analysis of solar interference,point structured light can eliminate the influence of the far-distance stripe information which caused by the line structured light.Besides,background subtraction can realize the removal of the sun spot and background in the obstacle structured light image.After the actual scene test,the results show that this method can effectively solve the problem of solar interference.(2)In view of the interference problem of fog on the obstacle structure light image in the railway tunnel entrance scene,an image processing method combining the structured light atmosphere model and the improved Tarel algorithm was proposed.According to the analysis of the light source of the obstacle detection platform for structured light measurement technology,the structured light imaging model is obtained.The Tarel algorithm is improved by histogram analysis to achieve the demisting,which can retain the obstacle stripe information and remove the fog noise at the same time.The test results show that the restoration similarity of the algorithm can reach 99.69%,which can effectively solve the problem of fog interference on the structured light image.(3)In view of the interference problem of rainfall on the obstacle structure light image in the railway tunnel entrance scene,an interframe replication restoration method based on histogram statistical threshold was proposed.Through histogram statistical threshold,the rain spot caused by rain water in the obstacle structure light image can be removed,and interframe replication can be used to repair the stripe defect caused by rain.Compared with the classical image restoration algorithm,the test results show that the algorithm has the best visual effect and restoration similarity.(4)Aiming at the problem of fast and accurate extraction of laser stripe in obstacle image,an adaptive convolution algorithm for stripe centerline extraction was proposed.The structure information and pixel information of the stripe are used to generate the convolution template to realize the Gaussian distribution of the stripe cross section,and the gray centroid method with a higher running rate is used to extract the stripe center point.The test results show that the extraction algorithm can accurately obtain the center position of obstacle stripe and provide high-precision conversion information for spatial positioning.In this thesis,the influence analysis and processing of the obstacle structure light image of several weather facing the railway tunnel entrance scene was completed.According to the visual studio 2015 platform and RK3399 platform,the performance test of the algorithm is preliminarily implemented,and the results verify the feasibility of the obstacle detection method based on structured light measurement technology.