In the intensity and polarization image fusion method driven by the difference features, the number of image features is large, and there is information fusion redundancy among the features. In this paper, a method of measuring the feature difference between infrared intensity and the polarization image based on cosine value is presented. By extracting the feature image from the source image, separating luminance, texture, and edge features, the difficulty of feature selection is avoided. Meanwhile, the feature difference degree is defined and calculated based on the feature vectors extracted from the feature map. The difference between the images is quantified in this way. Finally, the rationality and effectiveness of the proposed method are verified through experiments.%现有差异特征驱动的红外光强与偏振图像融合方法中,特征的选择上数量多,存在信息冗余,通过提取图像特征再求差来获取差异特征不能充分表征图像间的差异.本文提出一种基于余弦值的红外光强与偏振图像特征差异度的度量方法.通过对源图像提取特征图,将亮度、纹理、边缘三类特征进行分离,避免了图像特征选择的困难;定义了特征差异度,从特征图提取得到特征向量,再计算其差异度,对图像的差异进行了量化;最后,通过实验验证了所提方法的合理性和有效性.
展开▼
