摘要
针对单一方法进行纹理图像分类时易受旋转、光照等干扰的情况,提出了一种结合颜色特征和纹理特征的共生纹理分类方法。将图像转换到HSV颜色空间后,对H通道使用SLIOP算法以及对S和V通道用CLBP算法提取特征,然后将各自提取到的特征进行串联共生,最后利用支持向量机对纹理图像进行分类。基于被广泛使用的纹理图像数据库,对提出方法与其他典型分类算法进行实验对比,分析表明在分类的准确率和计算效率上获得了较大提升。实验结果表明,提出了方法具有较强的旋转不变性、光照不变性以及抗噪性。
This paper proposed a co-occurrence texture classification method to improve the rotation invariance and illumination robustness by jointing the texture features and the color features.Firstly,the method transformed the color space to HSV space.Secondly,it utilized simplified local intensity order pattern and completed local binary pattern to extract the color features in the H channel and the texture features in both S channel and V channel respectively.Thirdly,it concatenated these features as the descriptors of a color texture image.At last,it adopted SVM for classification.Based on experiments applying to the popular image datasets,it could be conclude that the proposed method possessed better computational efficiency and identification precision than other classical algorithms.Experiment results prove that the proposed method possesses rotation invariance and illumination robustness when it is used classification of color texture images.
作者
李君伟
刘光帅
刘望华
陈晓文
Li Junwei;Liu Guangshuai;Liu Wanghua;Chen Xiaowen(School of Mechanical Engineering,Southwest Jiaotong University,Chengdu 610031,China)
出处
《计算机应用研究》
CSCD
北大核心
2020年第7期2185-2188,共4页
Application Research of Computers
基金
国家自然科学基金资助项目(51275431)
四川省科技支撑计划项目(2015GZ0200)。
关键词
纹理分类
特征提取
HSV颜色空间
共生
简化局部像素强度模式
完备局部二值模式
texture classification
feature extraction
HSV color space
co-occurrence
simplified local intensity order pattern(SLIOP)
completed local binary pattern(CLBP)
