THE IMAGE AND SOUND PROCESSING LAB OF POLITECNICO DI MILANO

Oct 29, 2020 | Interviews

A snapshot from the SEG2019 conference

Who we are

The Image and Sound Processing Lab (ISPL) of the Dipartimento di Elettronica, Informazione e Bioingegneria (DEIB) of the Politecnico di Milano has more than twenty years of experience on applying advanced signal processing and machine learning techniques to multidimensional data.

In the last five years, ISPL has extended the field of application of its expertise to geophysical data. This has been possible due to the deep knowledge on seismic data processing that is present at DEIB thanks to Prof. Fabio Rocca and by Prof. Giuseppe Drufuca.

For many years, indeed, Prof. Rocca and Prof. Drufuca have contributed in a very significant way to research, development and validation of innovative tools for seismic processing and depth imaging.  Mainly in collaboration with ENI, they developed a number of state-of-the-art technologies for velocity model building, anisotropic ray tracing, multiple prediction and removal, diffraction imaging and reflection tomography.

As a matter of fact, researchers at DEIB have a deep and diffuse knowledge on signal and big data processing, with a special expertise on the treatment of large-scale inverse problems.

 

What we do

Thanks to this consolidated knowledge on seismic problems, the peculiar expertise of ISPL on computer vision and information retrieval techniques has allowed ISPL to propose to the scientific community some of the first machine learning and deep learning techniques proposed for the processing and interpretation of geophysical data.

In this regard, we can mention the participation in the first SEG machine learning contest (prediction of facies from well logs) in which the ISPL team was awarded among the first three competitors and  introduced the feature engineering used by all the top teams.

Visit of the Dean of Politecnico at ISPL laboratory.

Currently the main research topics at ISPL are:

  • Seismic processing via convolutional neural networks and deep learning
  • Geophysical interpretation via convolutional neural networks and deep learning
  • Least squares reverse time migration (LS-RTM)
  • Advanced regularization techniques and cost functions for Full Waveform Inversion

 

Believing in knowledge sharing, we are participating in several deep learning challenges and publishing our research codes not subject to disclosure restrictions on GitHub:

https://github.com/polimi-ispl

Where we head

At ISPL, we believe that the new generation of geoscientists will face crucial challenges for the future of the earth and for the sustainable development of our society.

Digitalization, artificial intelligence, high performance computing and cross-fertilization of ideas from various fields will be key to addressing these challenges.

The ISPL group, as a geophysics research group within a state-of-the-art IT department, is ready to jump on the train of the geophysicists of the future.

In the light of a growing network of geophysicists, we will be glad to share our passion and to collaborate with academic and industrial researchers in Italy, and around the world.

 

Contact us!

http://ispl.deib.polimi.it

linkedin.com/company/polimi-ispl

twitter.com/IsplPolimi

Selected recent publications

  • Bestagini, F. Lombardi, M. Lualdi, F. Picetti, and S. Tubaro, “Landmine Detection Using Autoencoders on Multipolarization GPR Volumetric Data,” IEEE Transactions on Geoscience and Remote Sensing (TGRS), pp. 1-14, 2020.
  • Kong, V. Lipari, F. Picetti, P. Bestagini, and S. Tubaro, “Deep prior-based seismic data interpolation via multi-res U-net,” in Society of Exploration Geophysicists (SEG) Annual Meeting, 2020, pp. 3159-3163.
  • Picetti, V. Lipari, P. Bestagini, and S. Tubaro, “Seismic Image Processing through the Generative Adversarial Network,” Interpretation, vol. 7, 2019.
  • Fortini, J. Panizzardi, N. Bienati, and V. Lipari, “Reflection FWI with Exponential Signal Encoding,” in European Association of Geoscientists and Engineers (EAGE) Conference and Exhibition, 2019.
  • Lipari, F. Picetti, J. Panizzardi, N. Bienati, and S. Tubaro, “Approximate Least Squares RTM via matching filters and regularized inversion,” in Society of Exploration Geophysicists (SEG) Annual Meeting, 2019.
  • Mandelli, F. Borra, V. Lipari, P. Bestagini, A. Sarti, and S. Tubaro, “Seismic Data Interpolation Through Convolutional Autoencoder,” in Society of Exploration Geophysicists International Exposition and Annual Meeting (SEG), 2018.
  • Fortini, V. Lipari, N. Bienati, J. Panizzardi, and S. Tubaro, “Robust reflection Full Waveform Inversion with exponential signal encoding,” in Society of Exploration Geophysicists (SEG) Annual Meeting, 2018.
  • Bestagini, V. Lipari, and S. Tubaro, “A machine learning approach to facies classification using well logs,” in Society of Exploration Geophysicists International Exposition and Annual Meeting (SEG), 2017.
  • Mandelli, V. Lipari, C. Fortini, and S. Tubaro, “First Break Tomography with TV Regularization and Structural Constraints,” in European Association of Geoscientists and Engineers (EAGE) Conference and Exhibition, 2017.
  • Mandelli, V. Lipari, C. Fortini, and S. Tubaro, “An investigation of uncertainty in velocity model building problems,” in Society of Exploration Geophysicists (SEG) International Exposition and Annual Meeting, 2017.
  • Lipari, D. Urbano, E. Spadavecchia, J. Panizzardi, and N. Bienati, “Regularized tomographic inversion with geological constraints,” Geophysical Prospecting, vol. 65, iss. 1, pp. 305-315, 2017.