机器视觉轨道缺陷检测成像系统的研究

发布时间：2018-04-13 23:26

本文选题：钢轨表面缺陷 + 高质量图像　；参考：《南昌大学》2017年硕士论文

【摘要】：轨道检测能够为铁路检修提供缺陷数据,正成为消除铁路安全隐患的必要手段之一。基于机器视觉的轨道缺陷检测由于其速度快、精度高、可靠性高等优势,成为轨道检测研究的一个热点。由于机器视觉系统中采集的原始图像的稳定性和质量的优劣会直接影响到后期图像处理的结果,合适的轨道缺陷成像系统能够降低轨道外部环境对成像系统的影响,提高图像处理的速度、降低漏检率和勿检率,因此对轨道缺陷成像系统的研究至关重要。本文以提高机器视觉轨道检测中成像效果为目的,以提高图像质量为切入点,深入研究成像方式对表面缺陷的影响、轨道表面缺陷特性和机器视觉成像模型,并通过实验验证模型的结论。本文主要研究内容如下:以钢轨表面高质量图像获取为目标,建立表面缺陷检测光学成像模型;分析系统振动对图像质量的影响,并将理论推广,研究基于线阵CCD相机机器视觉系统对轨面缺陷检测的数学模型,分析缺陷深度与图像灰度、成像方式之间的关系以及光照角度与图像灰度、对比度之间的关系,然后得出在45°的光源照射角度下,图片质量较其他角度高的理论结论,通过实验证明的模型中的结论的正确性。针对轨面光带强反光对轨面缺陷掩盖的问题,将偏振光引入轨道缺陷检测中,建立偏振光的机器视觉检测的光强度数学模型,研究图像灰度值、对比度与偏振片夹角的关系,分析偏振光对强反射光的抑制作用。实验证明采用偏振光能够有效抑制光带强反光,轨道图像质量较原系统有所提高。为了从客观角度分析验证不同成像系统采集图片的对比度高低,并保留人眼对图像对比度判断的特性,引入人眼感兴趣区域和非感兴趣区域权重系数,提出一种基于HVS的无参考图像对比度评价方法,经实验验证其与人眼主观评价结果相关性更好;然后用该方法对不同成像系统采集的图片进行评价,客观评价其对比度的高低,表明成像效果的优劣。通过客观量化的评价证明了成像系统优化的效果。
[Abstract]:Track detection, which can provide defect data for railway inspection and repair, is becoming one of the necessary means to eliminate the hidden trouble of railway safety.Track defect detection based on machine vision has become a hot research area because of its high speed, high accuracy and high reliability.Because the stability and quality of the original image collected in the machine vision system will directly affect the result of the later image processing, the proper track defect imaging system can reduce the impact of the orbit external environment on the imaging system.It is very important to improve the speed of image processing and to reduce the rate of missing and unchecked image processing, so it is very important to study the imaging system of track defects.In order to improve the imaging effect of machine vision track detection and to improve the image quality, this paper deeply studies the influence of imaging mode on surface defects, the characteristics of track surface defects and the imaging model of machine vision.The conclusion of the model is verified by experiments.The main contents of this paper are as follows: aiming at obtaining high quality image of rail surface, the optical imaging model of surface defect detection is established, the influence of system vibration on image quality is analyzed, and the theory is generalized.This paper studies the mathematical model of track defect detection based on linear CCD camera machine vision system, analyzes the relationship between defect depth and image grayscale, imaging method, illumination angle and image grayscale and contrast.Then the theoretical conclusion that the image quality is higher than other angles at 45 掳light source irradiation angle is obtained, and the correctness of the conclusion in the model is proved by experiment.Aiming at the problem that the rail surface defect is masked by the strong reflective light of the rail surface, the polarized light is introduced into the track defect detection, and the mathematical model of the optical intensity of the polarized light is established, and the relationship between the image gray value, the contrast and the angle between the polarizer and the optical intensity is studied.The inhibitory effect of polarized light on strongly reflected light is analyzed.The experiments show that the polarized light can effectively suppress the strong reflectance of the optical band, and the quality of the track image is improved compared with the original system.In order to analyze and verify the contrast of images collected by different imaging systems from an objective point of view, and to retain the characteristics of human eyes' judgment of image contrast, the weight coefficients of region of interest and region of non-interest are introduced.In this paper, a method of contrast evaluation without reference image based on HVS is proposed, which is proved to be more relevant to the subjective evaluation results of human eyes by experiments, and then the images collected by different imaging systems are evaluated by this method, and the contrast degree is evaluated objectively.The results show that the imaging effect is good and bad.The effect of the optimization of the imaging system is proved by objective and quantitative evaluation.
【学位授予单位】：南昌大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【相似文献】