一种新的纹理基元发现及表达方法

传统的纹理统计方法中对纹理基元采用形式语法的抽象表达使得算法非常复杂.基于图像关联规则挖掘,可发现纹理图像中的频繁模式,将频繁模式构建为纹理联合关联规则,纹理基元可以通过这些强关联规则特征矢量表达.实验证明,纹理关联规则纹理基元特征矢量能够很好区分纹理特征,并且易于表达以及在图像处理中使用,并具有方向不变性,能够为模糊...

A new representation method of image texton

Traditional texture statistical method is very complex because of the Texton representation method by formal syntax. Based on association rules data mining of image, the frequent patterns can be discovered and the textural combined association rules can be constructed by the frequent patterns. The texton can be represented by the characteristic vector of textural combined association rules. Experiments testified that the texton of characteristic vector of textural combined association rules can identify the textural features, and be easy to use in image processing. The texton has the characteristic of directional stability, which proposes a new idea for fuzzy and random texture image processing.