Machine vision detection device and method for steel rail surface discrete defects

Abstract

The invention discloses a machine vision detection device and method for steel rail surface discrete defects. The method includes the following steps that S1, a steel rail surface defect image is acquired; S2, a background smoothing filter is designed; S3, the original image is smoothened to obtain a background image; S4, the original image and the background image are differentiated; S5, dynamicthreshold value segmentation and connected region marking are conducted. According to the method, the smoothing filter is designed with local gray scale and gradient change information in the image asvariables, large-degree smoothing is conducted on the defect region with large gray scale and gradient changes in a local range, weak smoothing is conducted on shadows generated due to non-uniform reflection, rust and other interference regions, and a background image is generated, the original image and the background image are differentiated, differential image threshold segmentation is conducted, and steel rail surface defect detection is achieved. Influences of light changes, steel rail surface non-uniform reflection and rust can be effectively weakened while image defects are highlighted, and good steel rail surface defect detection effects are obtained under different rail environments.

Description

technical field [0001] The invention relates to the technical field of machine vision image processing, in particular to a machine vision detection device and method for discrete defects on the surface of a rail. Background technique [0002] With the development of high-speed and heavy-haul railways in China, strict requirements have been placed on the reliability of railway infrastructure. As an important part of railway track, steel rail will inevitably produce various contact fatigue defects on its surface under the action of running load and natural factors. The impact of the wheel running at high speed on the defective rail will cause the coupled vibration of the rail system, damage the line and vehicle components, reduce the ride comfort, and even endanger the driving safety. Therefore, timely detection of surface defects of rails and provision of maintenance data to railway departments are crucial to ensure the safety of railway transportation. [0003] At present,...

AI Technical Summary

Problems solved by technology

[0004] (1) The lighting conditions of different track environments and different time periods are complex and changeable, and changes in lighting conditions have a greater impact on detection performance;

[0005] (2) The rail surface is curved, affected by natural light, its surface will produce uneven reflection, resulting in shadow areas with low local gray values ​​in the collected rail images;

[0006] (3) The rails have been exposed to the field for a long time, and rust will appear on the surface under humid or corrosive conditions. On lines with low frequency of use, even large areas of corrosion will appear on the surface of the rails;

[0007] (4) The size and shape of discrete defects on the rail surface are different, and the distribution is not regular. It is difficult to detect by methods based on texture and shape features. Only local grayscale and gradient information are the most reliable features.

Images