基于支持向量机的路面图像分类方法

摘要

According to the characteristics of the pavement structure,the pavement classifica-tion and recognition method combined with color and texture features by using fuzzy support vector machine is proposed.The color moments of the pavement images in HSV color space are extracted as color features,and the gray-level co-occurrence matrix method is used to extract the texture features.Basing on the color features and texture features of pavement images,the features of the support vector are trained using fuzzy support vector machine.The characteris-tic vectors that satisfy the sample data characteristics of each image are obtained as much as possible by training.Through experiments,the accuracy of pavement classification and recog-nition by using the traditional support vector machine and fuzzy support vector machine are compared.The experiment results show that the proposed method is effective.%针对路面结构特征，提出一种颜色与纹理特征相融合并结合模糊支持向量机的路面分类识别方法。提取路面图像的 H SV颜色空间的颜色矩作为颜色特征，采用灰度共生矩阵法提取纹理特征，融合路面图像的颜色特征与纹理特征，采用模糊支持向量机进行支持向量特征训练，通过训练得到能尽可能多的满足每一种图像的样本数据特征的特征向量。通过实验，对比了采用传统的支持向量机与模糊支持向量机对路面分类识别的正确率。实验表明本研究所提出方法的有效性。