基于分形特征的人造目标探测方法研究

On the Research of Man-made Target Detection Beasd on Fractal Features

自然背景和人造目标对于某些分形特征存在着一定的本质差别，因此充分利用这些差别，能够为目标探测提供一套全新的方法。从仿真实验的处理结果来看，这种方法对于自然背景中嵌入少量人造目标这类简单情况的检测效果较好，且具有较强的抗噪声干扰能力；对于其他复杂情况，本文分别采用阈值分割组合法和K均值聚类法进行尝试，得到了一些初步结果。

Several fractal parameters have intrinsic chrenesbetween natural background and man-made targets. These differences present a new se of tools for target detection. For detecting man-made objects bo a natural background, fractalmethod is good.The results are still satisfactory in fow signalnoise ratio. In other comglexity cowditions, we get sevend initial results using threshold segmentation and K - mean clustering especially.