We will discuss recent advances in the use of optics to image behind walls. This is typically achieved by imaging and post-processing the return signals (light echoes) reflected from multiple surfaces, thus circumventing the obstacle or wall. The challenge therefore lies on the one hand in actually detecting the very weak return signals and on the other, in efficiently processing this data to retrieve target information for incoherent, multiple- scattered light. We will discuss the state of the art and show recent results demonstrating how deep learning can both simplify the detection hardware and achieve simultaneous location and identification of a person hidden behind a wall.

This presentation is part of Minisymposium “MS44 - 3D Image Depth/Texture/Reflectivity Tracking, Modelling and Reconstruction”
organized by: Catherine Higham (University of Glasgow) , Roderick Murray-Smith (University of Glasgow) .

Authors:

Daniele Faccio (Heriot Watt University)

Piergiorgio Caramazza (University of Glasgow)

Alessandro Boccolini (Heriot Watt University)

Daniel Buschek (University of Munich)

Matthias Hullin (University of Bonn)

Catherine Higham (University of Glasgow)

Robert Henderson (University of Edinburgh)

Roderick Murray-Smith (University of Glasgow)

Keywords:

deep learning, image representation, inverse problems