摘要
针对钢轨裂纹红外图像信噪比低、对比度低和纹理细节模糊的特点,提出了一种涉及降噪、目标细节增强和区域分割等的红外图像综合处理算法。首先,通过自适应均值滤波除去钢轨裂纹红外图像的部分噪声;其次,采用非线性非子采样Contourlet变换(Nonsubsampled Contourlet Transform,NSCT)对裂纹图像进行增强处理;最后,利用改进的最大类间方差法实现图像中目标裂纹区域的分割。实验结果表明,本文算法在有效地抑制噪声,提高图像对比度的基础上,能突显裂纹细节纹理,同时降低细节分割丢失,为后续图像信息处理奠定了必要的基础。
In response to the characteristics of low signal to noise ratio, low contrast and fuzzy texture of infra red thermal images of rail crack defect, an algorithm based on noise reduction, target detail enhancement and region segmentation was proposed to process the infrared images. Firstly, adaptive averaging filtering was a dopted to minimize part of the noise of rail crack infrared image. Then, the nonlinear NSCT (Nonsubsampled Contourlet Transform) was used to enhance the crack infrared image. Finally, a modified Otsu algorithm was applied to segment target crack region from the enhanced image. The experimental results indicate that the pro posed algorithm, in addition to effective noise suppression and image contrast improvement, can highlight tex ture details of the crack and reduce the loss of segmentation details, which lays a necessary foundation for the subsequent image information processing.
作者
顾桂梅
刘丽
贾文晶
GU Guimei;LIU Li;JIA Wenjing(School of Automation and Electrical Engineering,Lanzhou j iaotong University,Lanzhou 730070,China;Key Laboratory of Opt-Technology and Intelligent Control Ministry of Education,Lanzhou Jiaotong Univercity,Lanzhou 730070,China;Gansu Provincial Engineering Research Center for Artificial Intelligence and Graphics & Image Processing,Lanzhou 730070,China)
出处
《铁道学报》
EI
CAS
CSCD
北大核心
2018年第11期129-133,共5页
Journal of the China Railway Society
基金
甘肃省省级科技计划(1508RJZA059)