图像颜色与纹理特征的粗糙集分类模型

颜色与纹理作为两类基本特征,现有的很多图像检索、查询及分类系统都能很好地支持。该文提出一个基于图像内容的颜色与纹理特征,利用粗糙集分类的模型,即对图像进行四等分分割,而后对每一区域通过聚类分析得到4种主色,这样可构成16个颜色特征。再对聚类后的主色进行二值映射操作,并计算其0°与90°方向的共生矩阵,则可获取基于能量的8个纹理特征。在获取24个特征构成决策分类表后,最后利用粗糙集进行分类。实验证明其性能良好,相对于聚类、神经网络、贝叶斯等方法简单、高效、准确。

Rough Set's Classification Model of Image Based on Color and Texture Features

Color and texture as t wo basic features,in current image retrieval,query and classification system, it is supported well.This paper presents a rough set's classification model of image based on color and texture features.It imposes rough set classification m odel to segment image to quarter,clusters the segmented image to gain four ma in colors.It constitutes of16color features.Then it does binary mapping oper ation and computes the co-occurrence matrix of zero and ninety degrees.It cons titutes of8texture features,and applies the rough set to the classification ta ble based on the color and texture features.Experiment shows there is simplen ess,efficiency and nicety than some researches about image classification bas ed on cluster,neural network and bayesian algorithm.