PUPAx is a multidisciplinary project aiming to provide a new sophisticated product for Synthetic Aperture Radar (SAR)-based imagery that takes advantage of modern parallel computational means.
Left: input coherence layer of the city of Bam (Iran) after the 2003 earthquake. Center: input wrapper interferometric phase layer. Right: absolute phase obtained with our algorithm.
The target of PUPAx is to provide a faster software module for the computation of phase unwrapping, which is a complex key-algorithm widely used in SAR-imagery that requires up to several days of processing to converge to a solution. Achieving an efficient, state-of-the-art software implementation of the phase unwrapping algorithm on modern computational hardware, such as multicore and Graphics Processing Units (GPUs), is a challenging task, that needs a significant amount of knowledge and experience in the fields of computational mathematics, parallel computing, software engineering, and remote sensing. Through this project, we propose to fill the gap between the algorithmic maturity of phase unwrapping and the spread of massively-parallel devices, with a competitive new product that the remote sensing community demands.
Results using our parallel algorithm (orange and gray) compared to a graph-based sequential algorithm (blue).
Dr. Alessio Quaglino; Researcher; USI
State Secretariat for Education, Research and Innovation SERI Swiss Space Office grant nr. 236-01 D3;