【作者】 吴昌东；

【导师】 吴学杰；

【作者基本信息】 西南交通大学 ， 信号与信息处理， 2008， 硕士

【摘要】 本文以货车超限检测系统为研究背景,研究了图像法在超限检测系统中的应用。以数字图像处理技术为基础,重点介绍了小波相关理论在对货车图像的去噪及边缘检测中的应用。文章首先阐述了课题研究目的及意义,从实际的工程背景出发建立图像模型及确定技术方案。图像去噪方面,考虑到货车在运行过程中受特殊环境的影响,采集的货车图像含有椒盐噪声及高斯噪声等混合噪声。在小波传统阈值去噪的基础上,利用多级中值滤波可较好的抑制椒盐噪声并能较好保持图像细节而提升小波在抑制高斯噪声方面响应快等特点,本文采用了基于多级中值滤波与提升小波相结合的图像去噪算法,取得了比较好的实验效果,从而为有效地检测货车图像边缘打下好的基础。由于超限检测系统中采集的货车图像均是以天空为背景,本文采用小波变换模极大值边缘检测及边缘跟踪技术就能有效的检测出货车图像边缘;同时为了减少边缘检测所需时间,在去除伪边缘的阈值的选取上,采用改进型的最大方差自动取阈值法,不仅使阈值选取准确,而且很大程度上减少了阈值选取时间,从而降低边缘检测时间。实验表明:本文方法能有效的检测出货车边缘,为超限检测打下基础。文章最后简单介绍厂货车超限的相关基础知识,指出了判断货车超限方法及依据,分析了超限检测系统中图像处理部分所完成的功能。更多还原

【Abstract】 Based on the background of freight train gauge-exceeding detecting system, the thesis makes research on the image processing applied in the system. Based on the digital image processing,the thesis especially introduces the wavelet relevant theory applied in the freight train image denoising and the edge detecting.The thesis first presents the purpose and meaning, it builds the image module and decides the technology way from the practical project background. In the facet of image, According to the freight train will be influenced when it runs, so the collecting image includes the mix noise. Based on the common wavelet shrinkage, because the multistage median filter can not only denoise the salt and pepper noise but also keep the details of the image. The lifting wavelet can dismiss the time of denoising the gauss noise, so the thesis applys the two ways to deniose the mix noise. It gets good effect through practicing the image denoising by the thesis introduced way. So it can build the good foundation for the edge detecting.As the freight train image is collected by the background of the sky. The thesis applies the wavelet modulus maxima technique and the edge following technique to detect the freight train edge, which gets good outcome. According to the realistic and validity of the detecting, the thesis chooses the improved otsu way, it not only strongs the adaptive and accurate, but also dismisses the time of the choosed threshold and the edge detecting .The practices demonstrates that the way of the thesis can detecting the freight train edge, it builds the good foundation for the train gauge-exceeding detecting.In the end of the thesis, it introduces the related fundamental theories of the freight train gauge-exceeding detecting system, points to how to judge the freight train exceeding the gauge and analyzes the function of the image processing in the freight train gauge-exceeding detecting system.更多还原