11:00 AM - 1:00 PM
[STT39-P04] Noise removal of InSAR surface displacement images using Noise2Noise deep learning
Keywords:Noise2Noise deep learning, InSAR analysis, ALOS-2/PALSAR-2
Observing surface displacement is important for monitoring subsidence caused by Earth resource development. InSAR analysis estimates surface displacement from the phase difference of microwave observed by a satellite. However, the estimated displacement by InSAR analysis contains error components such as the phase delay in troposphere and decorrelation noise caused by scattering of microwave on the ground surface. Completely removing these noises with existing methods is sometimes difficult. Recently, a new method called Noise2Noise deep learning has been proposed in the field of image analysis. Unlike conventional deep learning, this method uses only noisy images to estimate noise distribution of images, and it has been shown that this method enables to remove noise in a high accuracy when using photographs. In this study, we examined the effectiveness of the Noise2Noise deep learning for the noise removal of InSAR surface displacement images using pseudo and real images.
First, we constructed pseudo noisy images by adding tropospheric noise and decorrelation noise to surface displacement. We assumed a circular subsidence for surface displacement, and used a mathematical model of fractal surface for tropospheric noise. For decorrelation noise, a random noise followed by Gaussian distribution was used. We then applied Noise2Noise deep learning to the constructed pseudo images. In the analysis using pseudo images, we changed the number of images used in the learning and compared the accuracy of noise removal using peak signal to noise ratio and root mean square error (RMSE). Subsequently, we applied the Noise2Noise deep learning to the ALOS-2/PALSAR-2 images taken for Kujukuri, Chiba, where the subsidence is occurring because of groundwater pumping.
In the analysis of 3000 pseudo images, the RMSE between image without noise and processed image for noise removing was 4.12 mm. In comparison with the previous study, we consider that the RMSE value of this study suggests that both tropospheric noise and decorrelation noise were successfully removed. Application of this noise removal process to the ALOS-2/PALSAR-2 images further confirmed its effects, however the noise reduction was not as significant as the analysis using the pseudo images. We will further improve the approach by changing some conditions or analyzing the noise distribution in real InSAR images. In conclusion, this study demonstrated the effectiveness of the Noise2Noise deep learning for the InSAR analysis mainly from the result of pseudo images.
First, we constructed pseudo noisy images by adding tropospheric noise and decorrelation noise to surface displacement. We assumed a circular subsidence for surface displacement, and used a mathematical model of fractal surface for tropospheric noise. For decorrelation noise, a random noise followed by Gaussian distribution was used. We then applied Noise2Noise deep learning to the constructed pseudo images. In the analysis using pseudo images, we changed the number of images used in the learning and compared the accuracy of noise removal using peak signal to noise ratio and root mean square error (RMSE). Subsequently, we applied the Noise2Noise deep learning to the ALOS-2/PALSAR-2 images taken for Kujukuri, Chiba, where the subsidence is occurring because of groundwater pumping.
In the analysis of 3000 pseudo images, the RMSE between image without noise and processed image for noise removing was 4.12 mm. In comparison with the previous study, we consider that the RMSE value of this study suggests that both tropospheric noise and decorrelation noise were successfully removed. Application of this noise removal process to the ALOS-2/PALSAR-2 images further confirmed its effects, however the noise reduction was not as significant as the analysis using the pseudo images. We will further improve the approach by changing some conditions or analyzing the noise distribution in real InSAR images. In conclusion, this study demonstrated the effectiveness of the Noise2Noise deep learning for the InSAR analysis mainly from the result of pseudo images.