SAR图像目标的方向梯度能量分形特征研究

Directed Gradient Power Fractal Features of SAR Image Objects

本文提出一种方向梯度能量分形特征提取方法用于目标的特征描述。该方法中,提出了方向梯度能量、方向梯度总能量、总方向梯度能量、方向梯度能量分形特征、复数分形特征等的概念和算法,分析并得出了方向梯度能量分形特征的性质,提取了二维目标的分形特征用于SAR图像目标检测。理论分析和实验结果表明,采用这种方法能够有效地检测不同形状和不同尺寸的目标。同时,这种方法还具有编程简单、运算速度快等优点。

A method, called directed gradient power fractal feature extraction, was presented for object representation. In the method, we proposed the definitions and algorithms of directed gradient power, directed gradient total power, total directed gradient power, directed gradient power fractal feature, and complex number fractal feature. The properties of directed gradient power fractal features were analyzed and concluded. Two-dimensional fractal features were extracted for SAR image object detection. Theoretical analysis and experimental results show that by using the method, the objects of different shapes and scales can be detected efficiently. The method has the advantages of simple programming, faster computation, etc.