红外目标识别图像复杂度度量方法综述

Survey of Image Complexity Metrics for Infrared Target Recognition

红外图像复杂度度量方法不仅可以用于描述目标识别面临的复杂场景变化,而且在红外成像系统性能预测与评估、目标识别算法性能对比、建立和改进目标获取性能模型等方面也有广泛而重要的应用。给定了红外目标识别图像复杂度的定义,对该领域近年来最新出现的和部分经典的度量方法进行系统的归纳总结和对比分析,提出了度量方法选择的依据,指出了现有度量方法的缺点和不足,并指出红外目标识别图像复杂度度量未来将向着融合多种特征或者综合多种度量方法的趋势发展。

Metrics of infrared image complexity were not only used to characterize the variation of complex scenario for target recognition, but also are applied widely in infrared imaging system performance prediction and evaluation, performance comparison of target recognition algorithms, constructing and improving target acquisition model, and other fields. The definition of image complexity for infrared target recognition is presented, and some recently presented and some classical image complexity metrics in this field are summarized and analyzed comparatively. The basis for choosing metrics is proposed, and the shortcomings and deficiencies of existing metrics are pointed out. Image complexity metrics for infrared target recognition will develop towards fusing many features or integrating many metrics in the future.