一种半调图像类型识别方法  

Recognition algorithm for halftone image classification

作　　者：孔月萍[1] 杜旭苗[1] 

机构地区：[1]西安建筑科技大学信息与控制工程学院,西安710055

出　　处：《计算机应用研究》2009年第12期4850-4851,共2页Application Research of Computers

基　　金：陕西省自然科学基金资助项目(2004F32);陕西省教育厅专项计划资助项目(04JK244)

摘　　要：半调图像类型的先验知识对于逆半调处理有着十分重要的意义。为了对半调图像进行类型识别,利用不同半调图像表现出的纹理差异设计了一种半调图像类型识别方法。其中主要选取了基于灰度—梯度共生矩阵的大梯度和小梯度优势统计值作为判别特征。仿真实验结果表明,所选取的特征值在不同类型半调图像上表现出明显的差异,将其作为半调图像分类识别的依据,最终通过判别函数可将常见的半调图像分为三大类。It is import to get the transcendental knowledge of the haltoning for the process of inverse hahoning. In order to classify the halftone images, designed a recognition algorithm, based on the texture differences among halftone images. The texture eigenvectors of the big-gradient preponderance and the small-gradient preponderance that could be produced by gray level-gradient co-occurrence matrix, were chosen as the discriminable scalars in the algorithm. Experimental results show that the eigenvalues have obviously diversity on different halftone images and the familiar halftone images can be classified into three classes by the algorithm.

关 键 词：图像分类 半调图像纹理 灰度—梯度共生矩阵 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]