基于多层次概念格的图像场景语义分类方法

A Semantic Classification Method of Image Scene Based on Concept Lattice Hierarchy

现有理论与方法在处理图像场景语义分类时,由于缺少对图像语义关系的深入挖掘,过分依赖视觉词典数量等原因,导致场景分类精度不高.本文提出一种基于多层次概念格的图像场景语义分类方法,将特征集转换成图像视觉形式背景,利用概念格的层次分类模型,通过层次映射关系,分别构建图像与视觉词集属性概念格,在此基础上通过动态调整阈值参数,获取分类精度概念外延阈值,得到具有较高分类精度的场景语义视觉模型.实验结果表明,该模型在精确度指标上有所提高,文中方法切实有效.

When dealing with the classification of image scene, due to the lack of in-depth mining of image semantic relations and reliance on visual dictionary, accuracy of current theory and method is not high. To solve the problem, a semantic classification method based on concept lattice Hierarchy is proposed. First, by using the hierarchical classification model of concept lattice, the feature set is transformed into the background of image visual form. Then, the concepts of image and set attributes of visual words are constructed respectively, based on the hierarchical mapping. Finally, a visual model with high classification accuracy is obtained by adjusting the threshold value dynamically. Experimental results show that the method is effective.