Detection of Internal Rail Defects Based on the Smoothed Randon Transform in The Presence of Ultrasonic Grain Noise
Most ultrasonic inspection systems rely on ASCAN mode to detect rail crack. The decision of the presence of a defect is then made based on a threshold set by the inspection operator. However, this operation is made more difficult when there exists ultrasonic and electronic noise, which leads to the false alarms or missing detection. The aim of this research work is to detect internal rail cracks in ultrasonic BSCAN mode images using a modified Radon transform, which is able to detect segment lines by considering a finite uncertainty on the line orientation and thickness in the presence of interfering ultrasonic clutter. The originality of this transform resides on the fact that it uses different mask functions to characterize the neighbouring region of the line pixels in the BSCAN image. In this paper, we present two main functions, the Gaussian cone mask and the Gaussian band mask. A Major advantage of this new technique is its on-line implementation thanks to the fast computational of the transform in function of a particular angle only which mainly related to the direction of the incident ultrasonic beam. In this paper, we illustrate the performance of our method over standard methods by carrying out a set of tests using simulated BSCAN images with simulated cracks and noise based on phenomenological characteristics as well by using images obtained from the simulation of physical clutter model. The validation has been also carried out on real BSCAN images obtained from a real ultrasonic inspection bench mounted on the SNCF train and results demonstrate the good behaviour of the new detector compared to the usual one. Finally, we present a graphical user interface under Matlab, which automates the simulation, analysis and detection of our approach.
Smoothed Radon Transform BSCAN grain noise detection crack Gaussian mask.
Mounir DJEDDI Patrice AKNIN
The French National Institute for Transport and Safety Research 2 avenue du général malleret-joinville f-94114 arcueil - France
国际会议
第十七届世界无损检测会议(17th World Conference on Nondestructive Testing)
上海
英文
491-498
2008-10-25（万方平台首次上网日期，不代表论文的发表时间）