Noise Parameter Estimation Two-Stage Network for Single Infrared Dim Small Target Image Destriping

The existing nonuniformity correction methods generally have the defects of image blur, artifacts, over-smoothing, and nonuniform residuals. It is difficult for these to meet requirements enhancement in various complex application scenarios. In particular, when are applied dim small target images, they may remove targets as noise points due over-smoothing. This paper draws on idea a residual network proposes two-stage learning based imaging mechanism an infrared line-scan system. We adopt multi-scale feature extraction unit design gain sub-network offset sub-network, respectively. Then, we pre-train two sub-networks independently. Finally, cascade into train it. experimental results show that PSNR our method can reach more than 15 dB, it achieve excellent performance different backgrounds intensities noise. Moreover, avoid losing texture details or after effectively removing