高光谱图像具有丰富的连续光谱信息,覆盖从可见光到红外波长范围内数百个波段,其数据特性使得在图像处理中有挖掘光谱内在特征的独特优势,从而有利于充分利用空间和光谱信息,检测感兴趣区域中的目标。但是由于高光谱数据的高维性、实际场景的复杂性以及有标签样本数量的有限性,高光谱异常检测仍面临背景和异常不易区分和检测精度低的问题。为了解决以上问题,本文提出一种面向高光谱异常检测的背景记忆模型。首先,通过无监督的基于密度估计的粗检方法,获得伪背景和伪异常向量。其次,设计基于异常突出正则项约束的背景记忆生成对抗网络模型,扩大伪背景和伪异常间的距离,并采用弱监督-伪标签的训练方式,使得网络具有较强的背景生成能力,同时弱化其对于异常的重构效果,减少对于背景和异常重构的泛化性,增强输出检测结果中背景和异常的差异和辨别度。此外,在特征域和图像域进行对抗学习,提高样本生成能力,更好地学习到输入样本的分布情况,提高网络对背景的生成能力。最后,使用非线性背景抑制方法减小虚警率,进一步提升检测精度。实验结果表明,本文方法相比其他检测算法在不同数据集上具有更好的检测效果。

Hyperspectral images have a wealth of continuous spectrum information, covering hundreds of bands from visible light to infrared wavelengths. The data characteristics of HSI make it a unique advantage to tap the inherent attributes of the spectrum in image processing, which is conducive to making full use of spatial and spectral information and detecting targets in the region of interest. However, due to the high-dimensionality of hyperspectral data, the complexity of actual scenes, and the limited number of labeled samples, hyperspectral anomaly detection faces the problem of indistinct background and anomalies and low detection accuracy. Therefore, we propose a background memory model for hyperspectral anomaly detection. Firstly, the pseudo background and anomaly vectors are obtained through an unsupervised rough inspection method based on density estimation. Secondly, we design a background memory generation adversarial network model based on anomalous prominent regular term constraints. What"s more, we expand the distance between the false background and false anomalies in a weak supervision-pseudo-label manner. Thus, the network has a strong background generation ability while the effect on anomaly reconstruction is weakened, which reduces the generalization of background and anomaly reconstruction and enhances the difference and discrimination between background and anomaly. Besides, we perform adversarial learning in the feature domain and image domain to improve sample generation ability to learn the distribution of input samples better and strengthen the ability to generate background. Finally, a nonlinear background suppression method is introduced to reduce the false alarm rate and further improve the detection accuracy. The experimental results show that our model has a better detection effect on different data sets than other detection algorithms.