路面纹理的多重分形特征描述与识别方法

Approach of multifractal feature description and recognition for pavement texture

应用多重分形理论验证了路面纹理的多尺度特性,研究了路面纹理多重分形谱的直接计算方法以及权重因子对计算结果的影响规律。测量了4组不同级配的沥青混凝土样块纹理高程值,利用小波变换消除噪声数据,分别计算各组样块纹理的多重分形谱参数,构建了一种应用多重分形谱描述与识别路面纹理特征的新方法。分析结果表明:路面纹理呈现出复杂性与自相似性特征,不同路面纹理之间的多重分形谱存在明显的差异性;路面纹理越复杂,高程波度越剧烈,多重分形谱参数越大,反之则越小;对于相同类型的路面纹理,多重分形谱形状和参数均相近。可见,多重分形是一种描述与识别路面纹理特征的有效方法。

Multifractal theory was applied to justify the multi-scale characteristic of pavement texture. The direct computational approach of multifractal spectra was explored, and the effect of weighted factor on multifractal spectra computational result was studied. Four types of asphalt sample pavement texture heights were measured, and eliminating noises were processed by using wavelet transform. The parameters of mulifractal spectra for each sample pavement texture were computed, and a novel approach to describe and characterize pavement texture feature was proposed based on multifractal spectra. Analysis result shows that pavement texture has the features of complexity and self-similarity, and the multifractal spectra of different pavement textures obviously vary. The more complex the pavement texture is, the more fluctuant the heights are, the larger the parameter values of multifractal spectra are, and vice versa. The same types of pavement textures have approximate multifractal shapes and parameters. So multifractal is an effective approach to describe and characterize pavement texture feature. 1 tab, 9 figs, 19 refs.