Font Size: a A A

Fast Segmentation Of Railway Track Image Based On Embedded System

Posted on:2018-03-23Degree:MasterType:Thesis
Country:ChinaCandidate:W B ZhangFull Text:PDF
GTID:2392330596956519Subject:Vehicle Engineering
Abstract/Summary:PDF Full Text Request
With the rapid development of rail transit lines,the operational security of transportation is becoming more and more important.Running safety has become a key problem in transportation work.The track state detection has become an important means to ensure the safety of the train.Image segmentation is an important basic processing of image recognition and image analysis,which plays an important role in the whole study.Embedded with the advantages of small size,low power consumption,simple structure,high reliability and high performance-cost ratio has been widely used,the utilization of embedded system to process track images accurately and rapidly which extracts the important areas,such as rail,fastener,sleeper and other areas,lays a good foundation for the subsequent rail fracture detection,fastener positioning analysis,sleeper defect treatment,thus to research the fast segmentation method of track image has an important significance for track detection.This article reads as follows:In this thesis,based on the contour feature of track images,choosing the theory of binary morphology to segment the track images.First,the comparative analysis experiments including all stages of denoising,enhancement and edge detection of image preprocessing are compared.The images added Salt and Gauss noise are filtered by the means of simulation,the peak signal to noise ratio and the algorithm running time are regard as the comparison standard to evaluate the effect before and after the filtering,as the result that median filtering is chosen from the filtering algorithms of mean,median,Wiener and SNN;Image contrast is enhanced by gray histogram equalization,and the importance of equalization is verified by contrast experiments;On the basis of increasing the gradient operator,optimizing and verifying the anti-noise performance of gradient operator on differential calculation,and demonstrating the excellent processing effect of the improved Sobel operator;Based on the preprocessing,filtering structural elements with large coverage and reasonable size in multi-direction are selected as the morphology operators by comparison test;The prominent anti-noise performance and filtering effect of the cascaded filter with morphological transform sequence are studied and verified.The above processes are proved by Matlab.In the realization of algorithm,the functions of the program transplantation,peripherals drive and multi task scheduling management based on the Linux kernel operating system are studied,the application environment of OpenCV on embedded Linux system platform is built,and QT is connected to the OpenCV function library,the software of track image segmentation algorithm adapt to embedded system is written,the hardware circuits of the railway inspection car acquiring and processing images based on embedded system are designed,the important circuits are expounded,the results of image segmentation are analyzed and the practicability of the algorithm is verified.
Keywords/Search Tags:track detection, image processing, fast segmentation, feature extraction, embedded system, OpenCV
PDF Full Text Request
Related items