基于空时域级联滤波的红外焦平面条状噪声消除算法

Destriping Method of Infrared Images Based on a Concatenated Spatial-temporal Filter

针对红外图像中的条状噪声问题,本文提出了一种基于空时级联滤波的条状噪声消除方法。在红外焦平面阵列中,每列设有一个读出电路用于读取该列传感器的响应。读出电路偏置电压的差异是红外图像中条状噪声的主要成因。根据红外辐射的空间相关性假设,本文利用空域非线性滤波从红外图像中提取偏置电压;然后,采用高斯-马尔科夫过程建立偏置电压随时间的变化的数学模型,并采用基于时域卡尔曼滤波的最小均方误差估计(MMSE)方法估计偏置电压;最后,从红外图像中减去偏置电压以消除红外图像中的条状噪声。为了验证所提算法的性能,分别利用仿真数据和实际红外数据进行实验,实验结果表明本文所提算法能够有效消除红外图像中的条纹噪声,并且不会造成图像细节的损失。

This paper introduces a spatial-temporal filter to eliminate stripe noise in infrared images.Detectors in the same column of the infrared focus plane array are read out by one readout circuit.The difference between voltage biases of the readout circuits results in strong stripe noise in infrared images.This paper proposes a method to estimate the voltage biases from the infrared images by using a spatial nonlinear filter based on the assumption of spatial correlation of infrared radiation.Meanwhile,the variation in the voltage via time is modeled by the Gaussian-Markov process,and the minimum mean square error estimation of the voltage biases is derived based on the Kalman filter in the temporal domain.The performance of the proposed method is studied using simulation data and real infrared data.The experimental results show that the proposed method can eliminate stripe noise and does not lose the details of the infrared images.