综合相似性在红外目标隐身效果评估中的应用

对于单张红外图像进行隐身效果评估，需要考虑像素点间反映的图像相似特征信息，而单一的相似性度量方法不能够全面准确地反映红外图像间相似性。在对单张红外图像均等分块的基础上，综合考虑了基于图像灰度直方图法、方向梯度直方图特征法、结构相似度法和目标分类4种相似性度量方法的优点，利用主成分分析法确定不同相似性度量方法的权重值，提出了基于综合相似性度量的评估方法。通过各相似性度量方法间横向与纵向的对比，分析不同遮挡情况下目标与背景图像间每种相似性度量方法的平均值和标准差。结果表明:综合相似性度量能够更准确地反映图像间相似性信息，更有效地处理单张红外图像隐身效果评估问题。

Application of comprehensive similarity in the evaluation of infrared target stealth effect

For the stealth effect evaluation of a single infrared image, it is necessary to consider the image similar feature information reflected by the pixels in the image. A single similarity measurement method cannot fully and accurately reflect the similarity between infrared images. On the basis of equally dividing a single infrared image, the advantages of four similarity measurements based on image gray histogram method, direction gradient histogram feature method, structural similarity method and target classification were considered comprehensively. The principal component analysis method was used to determine the weight values of different similarity measurements, and an evaluation method based on comprehensive similarity measurement was proposed. Through the comparison of horizontal and vertical experiments between the similarity measurement methods, the mean and standard deviation of each similarity measurement between the target and the background image for different occlusion situations was analyzed. The results show that the comprehensive similarity measurment can more accurately reflect the similarity information between images, the problem of single-infrared image stealth effect evaluation is more effectively processed.