Detection And Identification Of Light For Automatic Train Protection Cabinet Based On Color Image Processing

High-speed EMU is the main transportation for people to travel.As a result,it is important to ensure the safe and stable operation of train maintenance.However,the maintenance process of EMU mainly relies on machine vision to identify faults.Long-time night work affects the judgement ability of manual maintenance.Moreover,it is time-consuming and inefficient.With the construction of high-speed railway and the increased distances,the pressure of maintenance personnel is increasing,too.At the same time,computer science has developed rapidly in recent years.Machine vision is widely used in military,medical and industrial fields.In order to reduce the workload of staff,improving work efficiency and ensure the safe and smooth operation of trains.This paper presents a detection and recognition technology for the indicator light of automatic train over speed protection cabinet based on color image processing.In this paper,according to the actual characteristics of ATP cabinet indicator,various color spaces and indicator recognition are studied respectively.The indicator light on the cabinet is smaller than the traffic light on the highway,the color of the center of the image indicator lamp is not pure,and the background of the cabinet is complex,which makes the detection of the indicator lamp more difficult.In the present,median filtering is adopted,which does not affect feature extraction.It not only can remove noise,but also most useful information has been fully preserved.In the conversion from RGB to HSV,component H is used to solve the problem of difficult color discrimination in RGB space,and then S component is used to extract the signal region and determine the approximate region of the signal lamp.Mathematical morphology eliminates unwanted points,and then uses the prior knowledge of pattern recognition to further verify the signal lamp to be tested.Based on color image processing technology,this automatic recognition and alarm system is designed.The modulus and miniaturized design concept is adopted in the automatic fault image detection system for the over-speed protection cabinet of EMU.The automatic image recognition technology is used to alarm the fault image hierarchically.For different parts,step by step thinning,image contrast and hierarchical prediction algorithm are adopted to improve the efficiency of automatic fault identification of EMU and meet the requirements of hierarchical alarm.