基于局部密度方差理论的高性能混凝土骨料均匀度定量评价
Quantitative evaluation on aggregate uniformity of high performance？？concrete based on local density variance theory

为科学高效地评价高性能混凝土（HPC）骨料均匀度，针对骨料均匀度评价定量化难的问题，提出一种基于局部密度方差理论结合计算机数字图像处理技术的方法，对骨料均匀度进行定量评价。先对采集的高清图像进行滤波去噪处理，采用边缘提取获得骨料形态，进行腐蚀、膨胀操作、开闭运算处理获取骨料分布图，再利用提出的局部密度方差计算公式对随机选取的2个测区进行均匀度计算。研究结果表明：通过MATLAB编程可以实现图像去噪和骨料外形特征提取，骨料所处的位置被加粗加黑，骨料与过渡界面处被锐化处理，骨料间噪点被去除，为均匀度计算分析提供支撑；基于局部密度方差计算公式，提出的3种局部密度计算指标[WTBX]（由标准局部边长l？？0得出的均匀度U？？1、U？？2和所有局部边长l取平均值得出的均匀度U？？3）可对测区骨料分布均匀度进行数值计算，避免了人为主观因素的影响；对测区图像进行0？？°？？~90？？°？？旋转模拟拍摄角度差异，对比分析两测区旋转后图像和原图均匀度评价指标U？？1、U？？2、U？？3标准差约为1？？%？？，波动性较小，表明选用的3种指标评价的唯一性较好；U？？3在测区1和测区2的评价中方差仅为0.55？？%和0.48%，？？波动性较U？？1、U？？2小；[WT]随所取不同角度图像样本数量增加，3种均匀度指标趋于平稳，方差进一步降低，3种指标既可以单独评价，也可以综合评价HPC骨料的均匀性。
To solve the problem of scientific and highly efficient quantifying the aggregate uniformity of high？？performance concrete (HPC), a method based on density variance theory and computer digital image processing technology was proposed，and then the aggregate distribution of high？？performance concrete was quantified quantitative. The collected high？？resolution images were filtered and denoised. The aggregate shape was obtained using edge extraction, and the aggregate distribution image was obtained using corrosion, expansion, open and close operations. The calculation formula for local density variance was used to calculate the uniformity of the two randomly selected areas. The results show that the image denoising and the feature extraction methods for aggregate shapes can be realized by using MATLAB programming. In the image, the aggregate is thickened and blackened, the aggregate and the transition interface are sharpened, and the noise in the aggregate is removed, which can provide support for the analysis of uniformity. Based on the local density variance formula, three indices of the local density calculation for aggregate uniformity quantitative evaluation (i.e., ？？U？？？？1 and ？？U？？？？2, calculated at local side standard length ？？l？？？？0, and ？？U？？？？3, calculated using the average over the entire local side length ？？l？？) can be used to quantify aggregate uniformity, which can avoid the influence of artificial or subjective factors. The variance of the uniformity ？？U？？？？1, ？？U？？？？2, and ？？U？？？？3 for two selected HPC testing zones is approximately 1% after rotating 0° to 90° to simulate different shooting angles. The volatility is small, which indicates high consistency in the three indicators. The variance of ？？U？？？？3 for selected area 1 and area 2 are 0.55% and 0.48%, respectively; the volatility is lower than that in ？？U？？？？1 and ？？U？？？？2. For more samples taken from different angles, the three uniformity indices tend to be stable, and the variance value is