Signal Processing & Multimedia

Research activities:

The research activity covers several aspects of the image, video and 3D data processing, analysis and coding. In particular, the research group is active on several aspects of the acquisition, processing and coding of three-dimensional representations.

  • Acquisition of dynamic 3D scenes for 3D video applications
  • Compression of 3D representations, with a particular focus on multi-view video and depth maps
  • Heterogeneous sensors data fusion for the acquisition of static and dynamic 3D scenes
  • Automatic registration of multiple 3D views of a single object
  • Reconstruction and decision algorithms for muon tomography
  • Gesture recognition from color and 3D data
  • Scene segmentation from depth and color data
  • Content-based retrieval in 3D models databases
  • Image, video and 3D video post processing for coding artifacts reduction
  • Robust 3D video transmission over unreliable channels
  • Interactive transmission of three-dimensional scenes compressed in a scalable way

Homepage: http://lttm.dei.unipd.it

People: Giancarlo Calvagno (Contact person), Simone Milani, Pietro Zanuttigh

  • Advanced video streaming techniques for peer-to-peer networks
  • Colour image processing
  • Multimedia forensics

Homepage: http://lttm.dei.unipd.it

People: Giancarlo Calvagno (Contact person), Simone Milani, Pietro Zanuttigh

  • Alternating direction method of multipliers (ADMM) approaches for non-convex distributed optimization problems. We are interested in identifying efficient methods to solve a (non convex) minimization problem in a fully decentralized way, that is in such a way that only local parameters exchanges between neighbor nodes or neighbor regions are allowed.
  • Distributed optimization in the smart grid scenario. With the objective of increasing the reliability and the efficiency of power distribution, smart grids have emerged as a promising paradigm to integrate electrical, control and communication technologies. In this context we are interested in identifying fully distributed solutions to the grid optimization problem, by considering a typical scenario where renewables as well as local energy storage (batteries, or electric vehicles) are present.
  • Distributed localization in a sensor network. Distributed localization algorithms that rely on message exchange and distance measurements with neighboring nodes only, and where only a small subset of the nodes are anchors for which the true position is known, is the challenging non-convex problem that we are trying to solve efficiently and reliably.

Homepage: dgt.dei.unipd.it

People: Tomaso Erseghe (contact person)

  • Runtime compression of biometric time series.
  • Efficient compression/transmission of data over wireless channels.
  • Motion analysis: gait analysis from smartphone and wristband motion data, user identification and pattern analysis for preventive healthcare applications.
  • Activity recognition and behavioral profiling from motion sensors.

Homepage: http://signet.dei.unipd.it/internet-of-things/

People: Michele Rossi (contact person)