自然背景中人造信息的检测算法综述

Summary of algorithms for detecting man-made information in natural background

对目前自然背景中人造信息的检测算法进行了综述。由于自然背景与人造信息在某些特征上存在固有差异,因此可以利用这些差别检测出自然背景中的人造信息。分别讨论了基于几何特征、偏振特征、分形特征和概率模型的人造目标检测算法,并对天然背景中人造自然的检测算法做了初步的探讨。通过研究发现,基于分形特征的人造信息检测算法是目前研究的热点,而基于DRF模型的人造信息检测算法由于其检测精度高、适应性良好,正成为新的研究发展趋势。

The common algorithms of detecting man-made information in natural background are summarized. Because there are some intrinsic differences in some features between natural background and man-made information, different methods can be utilized to detect man-made information in natural background. The methods based on geometry model, polarization model, fractal model and probabilistic model are discussed. The methods of detecting artificial nature in natural nature background are introduced. According to the research, the algorithm of detecting man-made information based on fractal model is the hotspot at present, while the DRF model will attract more attention with its high accuracy and flexibility.