【摘要】 针对复杂云层背景下的红外弱小目标检测问题,提出了一种新的基于非参数统计的云层边缘纹理描述方法,并利用该方法构造出针对云层边缘的陷波滤波器.由于云层对阳光的漫反射特性,它的边缘区域成像时呈现出稳定的逐级过渡的纹理特征,而利用非参数统计中的秩分析方法可以较好地描述这种过渡纹理.首先,定义理想云层边缘纹理的秩统计量.然后,获取分析图像与理想云层边缘纹理秩统计量的相关曲面.最后,对秩相关曲面进行变换,构造出针对云层边缘区域的陷波滤波图像.将陷波滤波图像与高通滤波图像相乘,实现对残留云层边缘的抑制.实验结果表明,在没有对背景噪声的统计分布做任何假设的情况下,利用该方法能够对背景中的强云边缘进行有效抑制.更多还原

【Abstract】 To detect dim and small target from infrared images,a novel method for cloud edge description with nonparametric statistics was proposed.A notch fiter for cloud edge was constructed by using this method.It is found that there is stable gray level transition in the cloud edge regions due to its diffuse reflection of sun light.The rank analysis method is very suitable to describe the transition texture.First,the rank of ideal cloud edge texture was defined.Then,the rank correlation surface was calculated between object image and the ideal cloud edge region.Finally,notch filtering image used for cloud edge suppression was gotten by the rank correlation surface transformation.Then the high pass image was multiplied by the notch filtering image to suppress the remaining parts of cloud edges.The experimental results show that the proposed method can efficiently suppress the heavy cloud edges without any hypothesis on the probability density distribution of the background noise.更多还原