论文部分内容阅读
目的为了解决复杂背景干扰下基于线性滤波异常检测算法无法有效区分复杂背景特征与异常目标特征,导致检测结果虚警率偏高的问题,提出一种面向复杂背景的遥感异常小目标仿生非线性滤波检测算法。方法受生物视觉系统利用不同属性信息挖掘高维特征机理的启发,该算法通过相关型非线性滤波器综合多波段光谱数据提取高维光谱变化特征作为异常目标检测检测依据,弥补线性滤波抗噪性能差,难于区分复杂背景特征与目标特征的缺点。结果仿真实验结果验证该算法在仿真数据及真实遥感数据的异常检测效果上有较大改善,在实现快速异常检测的同时提高了检测命中率。结论本文方法不涉及背景建模,计算复杂度低,具有较好的实时性与普适性。特别是对复杂背景下的小尺寸异常目标具有较好的检测效果。
In order to solve the problem that the linear filtering anomaly detection algorithm can not effectively distinguish the complex background features from the anomalous target features under complex background interference and lead to high false alarm rate of the detection results, a biomimic nonlinear filter Detection algorithm. The method is inspired by the fact that the biological visual system uses different attribute information to mine the high-dimensional feature mechanism. The algorithm extracts the high-dimensional spectral variation features from the comprehensive multi-band spectral data by using the correlated nonlinear filter as the basis for abnormal target detection and detection, compensates for the linear filtering anti-noise performance Poor, it is difficult to distinguish the complex background features and the shortcomings of the target features. Results Simulation results show that this algorithm has greatly improved the anomaly detection effect of simulation data and real remote sensing data, and has improved the detection hit rate while realizing rapid anomaly detection. Conclusion This method does not involve background modeling, computational complexity is low, with good real-time and universal. Especially for small size anomalous targets in complex background, it has a good detection effect.