基于单元信息熵的半色调图像有效子块提取

Extraction of Effective Sub-blocks of Halftone Images Based on Unit Entropy

针对数字半色调图像有效子块的提取,提出了图像的单元熵及熵矩阵构造的算法.首先,将半色调图像分割成若干子块,对每一个子块的每一个像素点计算熵,得出熵矩阵.然后,对熵矩阵取均值及方差.最后,根据信息理论可得,当熵的均值越大且方差越小时,越能有效地获得子块.实验表明,通过单元信息熵获取的有效子块,含有信息量大,纹理变化平缓,适合用于分类.

In allusion to the extraction of effective sub - blocks of digital halftone images, this article proposes the unit entropy of ima- ges and the algorithm for the entropy matrix construction. Firstly, the halftone image is divided into several sub - blocks, and then the entropy for each pixel point of each sub - block is calculated to obtain the entropy matrix. Secondly, the study calculates the mean and variance of entropy matrix. Lastly, according to the information theory, when the entropy of the mean is greater and variance is smal- ler, sub - blocks can be more effectively obtained. Experimental results show that effective sub - blocks acquired through the unit en- tropy contain a large amount of information and the texture changes smoothly, so they are suitable for the classification.