摘要
针对传统路面信息获取方法效率低、强度大的问题,提出了公路路面图像的纹理分析方法,利用标准布朗运动模型描述公路表面,用标准分数布朗运动矢量作为分类的分形特征,将公路路面图像分成不重叠的像素块,通过计算每个像素块的标准分数布朗运动矢量获得各像素块的特征矢量.应用k-近邻分类实现图像块的分类.完好和损坏路面像素块的分类实验结果显示该方法是有效的.
A method for texture analysis of pavement surfaces is presented. Normalized fractal Brownian motion model is applied to describe the pavement surfaces. The Normalized fractal Brownian motion vectors (NFBV) are taken to be fractal features for the classification. The image of pavement surfaces is divided into nonoverlapping pixel blocks, and the attribute vectors are obtained by computing the NFBV of each pixel block. By using k-means clustering classifier, the experimental results of classification among pixel blocks of healthy pavement and distress pavement show that the method is efficient.
出处
《哈尔滨工业大学学报》
EI
CAS
CSCD
北大核心
2005年第6期816-818,共3页
Journal of Harbin Institute of Technology
关键词
分形
分数布朗运动
公路路面损坏
Brownian movement
Classification (of information)
Fractals
Image analysis
Mathematical models
Surfaces
Textures
Vectors
