A Method for Selecting SAR Interferometric Pairs Based on Incremental Coherence Spectral Clustering

The coherence level and number of selected interferometric pairs are directly related to the interferometric synthetic aperture radar (InSAR) phase estimation accuracy. In the Multi-Channel InSAR and Multi-Temporal InSAR, it is essential to select high-coherence interferometric pairs and remove low-coherence ones from the massive SAR singlelook complex (SLC) image data. The existing selection method, which based on basic coherence spectral clustering, may become increasingly computationally intensive when processing real-time data. To make a trade-off between the calculation cost and selection accuracy, a novel SAR interferometric pairs selection method based on incremental coherence spectral clustering is proposed. Experimental results demonstrate that the proposed method, which involves selecting a representative SAR SLC image for each cluster rather than re-constructing the adjacency matrix and re-estimating the number of clusters, can yield similar interferometric pairs selection results at a reduced computational cost.