摘要
半调图像分类在逆半调过程中是非常关键的步骤。文献[1]提出的神经网络分类算法利用增强的一维相关性,可以对半调图像进行适当的分类,但分类精度不够高。在神经网络分类算法的基础上,通过计算图像的灰度共生矩阵,进而提取图像的纹理特征来对图像进行分类。实验表明,改进后的算法可提高分类的精度。
The classification of halftone image is a key step in the process of inverse halftone. The neural net classification Algorithm can classify the halftone images using the enhanced one - dimensional correlation of halftone images, but the precision is not very high. This paper based on the neural net classification algorithm abstract the textural features by calculate the gray level co - occurrence matrix of halftone images. Experimental results show that the improved algorithm can enhance the classification precision.
出处
《现代电子技术》
2008年第22期112-114,共3页
Modern Electronics Technique
基金
陕西省自然科学基金资助项目(2004F32)
关键词
半调图像
图像分类
纹理特征
灰度共生矩阵
halftone image
image classification
texture feature
gray level co - occurrence matrix
