基于分形维数的路面裂缝图像分割方法研究

发布时间：2018-03-07 05:09

本文选题：路面裂缝　切入点：图像分割　出处：《长安大学》2014年硕士论文　论文类型：学位论文

【摘要】：近些年来，我国的公路网逐渐地扩展，公路的发展已经日趋成熟和完善，公路的养护和管理成为最关键的工作。路面裂缝是路面破损的最初形式，也是关系道路质量的最重要的一个指标。传统的人工检测方式存在不精确、效率低、成本高、耗时长、影响交通、安全隐患等诸多劣势。快速、准确、高效的路面裂缝自动检测技术是现阶段道路破损检测的发展方向，其对公路交通事业的发展具有重要意义。分形几何学是一门新兴学科，却发展迅速。而随着计算机技术的发展，数字图像处理技术也得到了广泛的应用。本文对差分计盒维数法进行了改进，并结合数字图像处理技术，提出了一种基于改进差分计盒维数法的路面裂缝图像分割方法。首先将原路面裂缝图像灰度化，再根据水平和垂直方向的灰度投影曲线判断裂缝的类型，，包括横向裂缝、纵向裂缝和网状裂缝，依据裂缝类型对图像进行缩小，继而采用对数变换对图像进行增强，再使用自适应中值滤波法对图像去噪。然后用改进的差分计盒维数法计算图像的局部分维数，本文主要是采用动态覆盖网格的方法解决原有方法对图像尺寸限制的问题，并且用对不同尺度分组的思想提高算法运行的速度。由于分维数能够描述不同的自然物或纹理，最后用局部动态阈值法寻找局部分维数值的最优阈值，路面图像自然地被分成裂缝和背景两部分，至此完成路面裂缝图像的分割。最后分别选取水泥路面和沥青路面的裂缝图像进行实验，与经典的图像分割方法作对比，实验结果表明本文方法的分割结果较好，提取的裂缝位置较准确，抗噪性能较好。
[Abstract]:In recent years, the highway network of our country has gradually expanded, the development of highway has become more and more mature and perfect, the maintenance and management of highway has become the most critical work. Pavement crack is the initial form of pavement damage. It is also one of the most important indicators of road quality. Traditional manual detection methods are imprecise, low efficiency, high cost, time-consuming, affect traffic, safety hidden dangers and many other disadvantages. Efficient automatic detection technology of pavement cracks is the development direction of road damage detection at present, which is of great significance to the development of highway traffic. Fractal geometry is a new subject, but it develops rapidly. With the development of computer technology, digital image processing technology has been widely used. In this paper, the difference box-counting dimension method is improved. Combined with digital image processing technology, a method of pavement crack image segmentation based on improved difference box-counting dimension method is proposed. Firstly, the original pavement crack image is grayscale. Then according to the horizontal and vertical gray projection curve to judge the types of cracks, including transverse cracks, longitudinal cracks and mesh cracks, according to the type of cracks to reduce the image, and then use logarithmic transformation to enhance the image. Then the adaptive median filter is used to Denoise the image, and then the improved difference box-counting dimension method is used to calculate the local fractal dimension of the image. In this paper, the dynamic overlay mesh method is mainly used to solve the problem of the original method to limit the size of the image. The idea of grouping different scales is used to improve the speed of the algorithm. Because the fractal dimension can describe different natural objects or textures, the local dynamic threshold method is used to find the optimal threshold value of partial local dimension. The pavement image is naturally divided into two parts: crack and background, so that the segmentation of pavement crack image is completed. Finally, the crack images of cement pavement and asphalt pavement are selected for experiments, and compared with the classical image segmentation method. The experimental results show that the segmentation results of this method are better, the crack location extracted is more accurate, and the anti-noise performance is better.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP391.41;U418.6

【参考文献】