论文部分内容阅读
高光谱遥感目标检测是遥感信号处理领域的热点问题,基于核机器学习的KRX算法能充分利用高光谱波段间的非线性光谱特性,在原始光谱的特征空间进行探测,能够获得较好的检测效果。针对KRX算法检测过程计算复杂、不能满足快速处理要求的缺陷,引入了卡尔曼滤波器的递归思想,提出了一种核递归的高光谱异常目标检测算法。从光谱分析的角度,应用Woodbury引理从上一时刻的状态迭代更新当前像元的Gram核矩阵,避免了高维矩阵数据重复计算。实验结果表明,与传统RX、因果RX和KRX等算法相比,在检测精度有所提高的同时,大大缩短了算法检测时间,提高了异常目标检测效率。
Hyperspectral remote sensing target detection is a hot issue in the field of remote sensing signal processing. The KRX algorithm based on nuclear machine learning can make full use of the nonlinear spectral characteristics of hyperspectral bands to detect in the feature space of the original spectrum, and can obtain better detection results . Aiming at the defect that the calculation process of KRX algorithm is complex and can not meet the requirement of fast processing, a recursive idea of Kalman filter is introduced and a target recursive hyperspectral target detection algorithm is proposed. From the perspective of spectral analysis, Woodbury’s lemma is used to iteratively update the Gram kernel matrix of the current pixel from the state of the previous moment, avoiding the repeated computation of high-dimensional matrix data. Experimental results show that compared with the traditional algorithms such as RX, causal RX and KRX, the detection accuracy is greatly improved while the algorithm detection time is greatly shortened and the detection efficiency of abnormal targets is improved.