基于间隔采样的快速变分条带噪声检测方法

Fast variational detection of stripe noise based on interval sampling

为了提高目前主流卫星推扫成像过程中多列条带噪声的检测效率,提出了一种基于间隔采样的快速变分条带噪声检测方法。该方法以条带噪声成分变分建模和优化求解为基础,通过间隔采样和构建带间隔采样参数的条带噪声成分估计模型,完成条带噪声成分的快速求解,然后对条带噪声成分列均值进行一元离群点检测和后处理,完成条带噪声的定位。由于采用间隔采样的策略,该方法在不损失条带噪声检测精度的情况下显著提高了检测效率。

This study proposed a fast variational detection method for stripe noise based on interval sampling,aiming to improve the detection efficiency of multi-column strip noise during the pushbroom imaging of mainstream satellites.Based on the variational modeling of stripe noise components and the optimal solution,this method can quickly determine stripe noise components through interval sampling and establishing an estimation model of stripe noise components with interval sampling parameters.Then,this method can locate the stripe noise through the one-dimensional outlier detection and post-processing of the column mean values of stripe noise components.Owing to the interval sampling strategy,the method proposed in this study significantly improves the detection efficiency without impairing the stripe noise detection accuracy.