基于分形理论的红外图像机场跑道自动目标识别

针对红外机场跑道航空图像特点,提出一种新的基于纹理结构的机场目标自动识别方 法。根据红外图像中机场目标的灰度和分形几何特征,利用先假设后检验方法对感兴趣区域进行分割;然后利用形态学滤波完成机场跑道潜在目标的进一步分割;最后,根据机场跑道的几何特征,连接图像边缘点寻找机场跑道的两条平行直线;根据纹理图像直方图统计特征完成机场目标的自动识别。实验结果证明:该方法有较高的自动目标识别率和计算实时性。

Automatic Target Recognition of Airfield Runway in Infrared Images Based-on Fractal Theory

Aiming for the feature of the airfield runway in long range aerial photography infrared images,a novel threestep recognition strategy based on fractal theorywas presented. First, regions of interest (RO I) , are segmented based on fractal texture features according to supposition before verification. Then, latent airfield targets are performed over all sub imageswith morphological filters. Finally, the airfield target is recognized automatically according to the geometric characters and histogram statistic matching based on texture features. Experimental results on the real images demonstrated that the approach performs well in automatic recognition accuracy and computational efficiency.