摘要
鉴于红外装甲目标检测中红外图像对比度低、背景复杂,导致图像的信噪比低而难于进行目标的检测,提出一种基于小波和改进的分形理论相结合的背景抑制方法。针对红外图像呈现的相关性强的特点,利用小波分析将图像中的低频缓变背景滤除,得到包含目标和强边缘杂波的图像;又由于目标分形维数对尺度的敏感程度高于边缘杂波的分形维数,提出通过计算图像在不同尺度内不规则因子的变化率来进一步抑制背景中的边缘杂波。实验表明:该算法能显著提高图像的信噪比(信噪比增益在2左右),对背景边缘有很好的抑制效果。
It is hard to detect the infrared armor target because the background is so complex and the SNR of image is low. Consequently, a new background suppression method is put for- ward which bases on the wavelet and improved fractal theory. Filtering the low frequency background in image by using wavelet transformation, image with the target and edge clutter can be obtained. Because fractal dimension of target is much more sensitive than that of the edge clutter, so the edge clutter in background can be inhibited by calculating the rate of irreg- ular factor of the image in different scale, so as to achieve a good background suppression effect. Experiment results show that this algorithm can significantly improve image signal-to- noise ratio （GSNR is about 2）, and has good inhibition effect on undulate background.
出处
《应用光学》
CAS
CSCD
北大核心
2013年第3期527-531,共5页
Journal of Applied Optics
关键词
背景抑制
小波
分形理论
多尺度不规则度
background suppression
wavelet
fractal theory
multi-scale irregularity
