基于分形曲面尺度斜率特征的弱小目标检测

Dim target detection based on the scale slope feature of fractal surface

鉴于弱小目标检测所固有的难点及常用的检测方法不能准确、稳定地检测出目标,提出了运用分形曲面尺度斜率特征检测弱小目标的方法。通过实际数据分析可以得出:相比常用的分形维数和分形拟合误差等检测特征,分形曲面尺度斜率特征在表征人造目标与自然背景的差异上更加明显,在抗图像噪声干扰上也更为优异,有着更强的鲁棒性。该方法普遍适用于检测自然环境中的弱小目标,尤其在对空弱小目标方面,检测概率更高。无论背景、飞行姿态、目标类型发生怎样的变化,经本文算法运算后只需一步简易的分割就可以检测出微弱暗小目标。

In view of the intrinsic difficulties of dim target detection and the shortcomings of common detection method,a new method of dim target detection based on scale slope feature of fractal surface was put forward. Compared with fractal features of fractal dimension and fractal fitting error,the scale slope feature of fractal surface is more obvious in representing the difference between target and background,and it is more robust than the traditional fractal features. This method is suitable for detecting dim targets in a natural environment,and it has higher detection probability especially for the small target in the sky. It is not affected by the target type,object pose,natural background etc.After fractal feature process,dim target can be detected through simple segmentation.