【摘要】 针对复杂背景下红外弱小目标图像背景抑制难题,提出了一种基于曲面拟合的双向扩散滤波红外背景抑制新算法。采用高斯Facet模型拟合邻域图像曲面,采用综合方向导数梯度(IDDG)算子描述拟合图像的灰度特征,进而对双向扩散滤波进行改进,并将其与IDDG算子相结合,发展出了具有解析形式的改进的双向扩散滤波算法,给出了该算法关键参数的自适应选取方法。与传统的背景抑制算法相比,本文算法对图像灰度特征的描述更准确,并能据此在前向扩散和后向扩散之间自适应地切换,从而实现了在抑制背景杂波的同时增强目标能量,且能够克服传统算法处理椒盐噪声方面的缺陷。理论分析与仿真实验表明,本文算法对包含强纹理结构的复杂背景杂波具有良好的抑制作用和稳健的适应作用,对于信噪比为0.8的图像,可获得21.6的信噪比增益。更多还原

【Abstract】 Aiming at the difficulty of infrared week-dim targets detection with strong background clutter,a novel infrared background suppression algorithm based on surface fitting bidirectional diffusion was proposed,the GaussianFacet model was applied to fit the underlying intensity surface of the infrared week-dim targets image,and integrated directional derivative gradient(IDDG) operator based on Gaussian-Facet model was used to describe the gray-scale character of the image.The analytical bidirectional diffusion filter is developed by combining IDDG operator with improved bidirectional diffusion filter,meanwhile,the key parameters of the algorithm are determined adaptively.Compared with traditional background suppression algorithms,our method describe the gray-scale character more accurately,and it can switch adaptively between forward(background suppression) and backward(targets enhancement) diffusion processes according to the gray-scale character of the image.Moreover,the algorithm overcomes the shortcomings of dealing with salt and pepper noise.Theoretical analyses and experimental results show that this method provide good filtering performance and adaptability for infrared week-dim targets with strong clutter background,the increase of the signal-to-noise ratio(ISNR) reaches 21.6 when the signal-to-noise ratio(SNR) of the image is 0.8.更多还原