Technologies and Architectures for Machine Perception - TAPEC

Reference of the Group:

GIUV2017-349

 
Description of research activity:
The TAPEC group researches on hardware devices and architectures for computer perception, with a special focus on image processing and vision devices. In the field of perception architectures, research is carried out on the use of FPGAs for image and visual event processing. In recent years, architectures have been developed that work with pixel or event streams instead of typical image processing. The most active field of research at present is the design and fabrication of event-driven vision cameras. The TAPEC group has developed a proprietary chip, in CMOS technology, that sends the pixel that has changed the most since it was last read, instead of sending the pixels in order as in conventional images. This technique has advantages such as selective reduction of the information to be processed, simple and synchronous interface, high processing speed and motion analysis, etc. As an example, the latest camera developed is capable of tracking objects with a temporal resolution of microseconds; to achieve the same with a conventional camera, it would have to operate at a rate of 500,000 images per second, with hardware capable of processing all that information in real time, which...The TAPEC group researches on hardware devices and architectures for computer perception, with a special focus on image processing and vision devices. In the field of perception architectures, research is carried out on the use of FPGAs for image and visual event processing. In recent years, architectures have been developed that work with pixel or event streams instead of typical image processing. The most active field of research at present is the design and fabrication of event-driven vision cameras. The TAPEC group has developed a proprietary chip, in CMOS technology, that sends the pixel that has changed the most since it was last read, instead of sending the pixels in order as in conventional images. This technique has advantages such as selective reduction of the information to be processed, simple and synchronous interface, high processing speed and motion analysis, etc. As an example, the latest camera developed is capable of tracking objects with a temporal resolution of microseconds; to achieve the same with a conventional camera, it would have to operate at a rate of 500,000 images per second, with hardware capable of processing all that information in real time, which is not feasible today.
[Read more][Hide]
 
Web:

http://tapec.uv.es

 
Scientific-technical goals:
  • Realizar sensores de vision selectiva guiados por cambios
  • Desarrollar arquitecturas para el procesado de flujos de pixeles para el analisis de movimiento
 
Research lines:
  • CMOS vision sensors.Design of integrated circuits for vision, especially application-specific selective gear-guided vision and cameras.
  • Specific architectures and FPGAs.Design of application-specific architectures using FPGA. Specific computer vision architectures.
  • Robotics and control.Design of robots, vehicles, sensorisation. Design of closed-loop control architectures.
 
Group members:
Name Nature of participation Entity Description
FERNANDO PARDO CARPIODirectorUniversitat de València
Research team
JOSE ANTONIO BOLUDA GRAUMemberUniversitat de València
 
CNAE:
  • -
  • -
 
Associated structure:
  • Computer Science
 
Keywords:
  • Sensores de Visión, Circuitos integrados, CMOS
  • FPGA, programmable logic, vision architectures
  • Robotics, control, automatic