红外偏振和光强图像差异特征分类是融合算法随着差异特征类型的变化而自适应变化的前提。构建了差异特征分类树, 以此实现差异特征分类。首先分析红外偏振和光强成像的差异特性, 依据其成像差异特性构建分类树第 1层差异的类别; 然后对多组图像统计并描述第 1层差异类别下的各差异信息, 依据统计结果构建第 2层差异的类别; 最后提取红外偏振和光强图像的差异特征, 将其按照差异特征分类树进行分类。实验表明, 所建立的差异特征分类树可将红外偏振和光强图像的差异特征分类。

Difference features classification between infrared polarization and intensity images is a precondition for fusion algorithms changing along with the change of difference features. This paper presents a classification tree of difference features to classify the features. Firstly, the difference characteristic of infrared polarization and intensity imaging is analyzed, with the first layer of the classification tree based on their imaging difference characteristics built; then statistic and description of the difference information to multi-group images is got, below the first layer of each differences category; according to the results, the second layer of the classification tree is built; and finally difference features of infrared polarization and intensity images are extracted, which are classified according to classification tree of difference features. Experiments show that the classification tree can classify the difference features of images between infrared polarization and infrared intensity images effectively.