基于图像处理的沥青混合料油石比识别方法研究

发布时间：2018-01-14 20:25

本文关键词：基于图像处理的沥青混合料油石比识别方法研究　出处：《长沙理工大学》2014年硕士论文　论文类型：学位论文

【摘要】：沥青混合料油石比的检测是沥青路面施工质量控制和性能评价的重要检测手段,为此,针对常用的油石比检测方法存在费时费力、精度较低、污染环境和危害人体健康等问题,本文基于图像处理技术,对沥青混合料油石比和空隙率的图像识别方法进行研究。(1)选取集料、结合料和级配不同的5种沥青混合料：上面层1#SBS改性AC-13C、中面层2#SBS改性和3#普通AC-20C、下面层4#和5#普通AC-25C,采用马歇尔试验法分别进行了各沥青混合料的配合比设计,确定了相应的最佳油石比,分别为5.1%、4.4%、4.3%、3.6%和3.6%。通过车辙、冻融劈裂、浸水马歇尔、低温弯曲和构造深度等性能试验,验证了各沥青混合料配合比设计结果的合理性。(2)对各沥青混合料按不同油石比击实成型标准马歇尔试件,对试件进行水平切割,获取多个切面,并采用数码相机拍摄了沥青混合料各切面的多张数字照片；据此采用阈值分割法进行了图像分割,并经图像去噪、裁剪等处理,提取出以试件中心为中心的900×900像素的矩形沥青混合料切面图像。(3)对提取的图像进行灰度化处理,分析获得了切面图像灰度频率分布曲线。结果表明,灰度频率分布曲线具有典型的双峰性质,且这种“双峰一谷”的性质与混合料的种类无关；左峰半枝曲线随着沥青混合料油石比的增大,逐渐从尖削变成饱满,所跨的灰度值范围逐渐变大,而最左侧的低矮台阶随着油石比的增大逐渐消失。因此,左峰半枝曲线可以反映油石比的大小,而低矮台阶与空隙率大小有关。(4)通过拟合左半枝灰度频率曲线,确定了合理的沥青和空隙的灰度范围,分别计算5种沥青混合料不同油石比下多张图像的识别值。结果表明,灰度频率分布曲线的左峰半枝曲线可以视为加权的双正态分布概率密度函数曲线的叠加,两者间的复相关系数达到0.99；相应得到的识别沥青面积比及空隙面积比分别与实际油石比及实际空隙率之间均存在着显著的线性相关性,相关系数均超过0.9,但不同沥青混合料的线性关系不同；据此通过相关关系验证了图像识别结果,其与实际数据之间的绝对误差一般在±0.2%以内,表明图像识别结果具有较高的精度,满足工程应用的需要。综上所述,采用图像处理技术对不同沥青混合料的油石比和空隙率进行识别是可行的,可为工程应用提供一种快捷、简便的检测方法。
[Abstract]:The detection of asphalt mixture ratio is an important means of asphalt pavement construction quality control and performance evaluation. Based on the image processing technology, this paper studies the image recognition method of asphalt mixture oil / stone ratio and porosity. Five kinds of asphalt mixture with different binder and gradation: top layer 1 #SBS modified AC-13C, medium surface layer 2 #SBS modified and ordinary AC-20C. The mixture ratio of asphalt mixture is designed by Marshall test method, and the optimum ratio of asphalt to stone is determined by Marshall test method, which is 5.1% or 4.4% respectively. By rutting, freeze-thaw splitting, immersion Marshall, low temperature bending and structural depth tests. Verify the rationality of the design results of each asphalt mixture mix. 2) the asphalt mixture according to different asphalt stone compaction molding standard Marshall specimen, the specimen is cut horizontally to obtain a number of cutting planes. The digital camera is used to take many digital photographs of each section of asphalt mixture. Based on this, the threshold segmentation method is used to segment the image, and the image is de-noised, clipped and so on. A 900 脳 900 pixel rectangular bituminous mixture section image with the center of the specimen as the center is extracted. The gray frequency distribution curve of the section image is obtained. The results show that the gray level frequency distribution curve has the typical bimodal property, and the property of this kind of "double peak and valley" is independent of the kind of mixture. With the increase of bituminous asphalt mixture ratio, the left peak half branch curve gradually turns from sharp to full, and the gray value range of the left peak gradually becomes larger, while the leftmost low step gradually disappears with the increase of oil stone ratio. The left peak half branch curve can reflect the size of the stone ratio, while the low step is related to the porosity. 4) by fitting the left half branch gray scale frequency curve, the reasonable gray range of asphalt and voids is determined. The results show that the left peak half branch curve of the gray frequency distribution curve can be regarded as the superposition of the weighted double normal distribution probability density function curve. The complex correlation coefficient between the two was 0.99; There is a significant linear correlation between the identified asphalt area ratio and the voids ratio and the actual voids ratio, respectively, and the correlation coefficient is more than 0.9. But the linear relation of different asphalt mixture is different; The absolute error between the result and the actual data is generally within 卤0.2%, which indicates that the result of image recognition has a high accuracy. To meet the needs of engineering application, it is feasible to use image processing technology to identify the oil-stone ratio and voids of different asphalt mixtures, which can provide a quick and simple detection method for engineering application.
【学位授予单位】：长沙理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U414

【参考文献】