基于Kolmogorov复杂性的垃圾图像分类模型

摘要

To further fight against the wide spread of image spam, this paper first proposed a spam image classification model based on Kolmogorov complexity.It used data compression to classify spam images effectively.Compared with current mainstream classification methods for spam image, the model needed neither text extraction from image, nor the feature definition and feature selection of image.It is a kind of parameter-free classification method.And the effectiveness and robustness of the model are verified through the experiments.Also Kolmogorov complexity is indicated to be promising in spam filtering.%为了进一步遏制图像型垃圾邮件的泛滥,首次提出了一种基于Kolmogorov复杂性的垃圾图像分类模型.该模型利用数据压缩技术,实现了对垃圾图像的有效分类.与目前主流垃圾图像分类方法相比,本模型既不需要提取图像中的文字,也不需要对图像特征进行定义和选择,而是一种无参数的分类方法.实验验证了本模型的有效性和鲁棒性,同时还表明,Kolmogorov复杂性在垃圾信息过滤中具有广阔的应用前景.