Deep learning networks for selection of measurement pixels in multi-temporal SAR interferometric processing

In multi-temporal SAR interferometry (MT-InSAR), persistent scatterer (PS) pixels are used to estimate geophysical parameters, essentially deformation. Conventionally, PS pixels are selected based on the estimated noise present in the spatially uncorrelated phase component along with look-angle error in a temporal interferometric stack. In this study, two deep learning architectures, namely convolutional neural network for interferometric semantic segmentation (CNN-ISS) and convolutional long short term memory network for interferometric semantic segmentation (CLSTM-ISS), based on learning spatial and spatio-temporal behaviour, respectively, were proposed for selection of PS pixels. These networks were trained to relate the interferometric phase history to its classification into phase stable (PS pixels) and phase unstable (non-PS pixels) measurement pixels using ~10,000 real world interferometric patch images of different study sites containing man-made objects, forests, vegetation, uncropped land, water bodies, and areas affected by lengthening, foreshortening, layover and shadowing. The networks were trained using training labels obtained from the Stanford method for Persistent Scatterer Interferometry (StaMPS) algorithm. However, pixel selection results, evaluated using a combination of R-index, Similar Time Series Interferometric Pixel (STIP) maps and a classified image of the test dataset, reveal that CLSTM-ISS estimates improved the classification of PS and non-PS pixels as compared to those of StaMPS and CNN-ISS. The predicted results show that CLSTM-ISS reached an accuracy of 93.50%, higher than that of CNN-ISS (89.21%). CLSTM-ISS also improved the density of reliable PS pixels compared to StaMPS and CNN-ISS. Further, the architecture outperformed StaMPS, and is expected to compete with other MT-InSAR algorithms in terms of computational efficiency.