Active Vision Sensor


In LASMEA-GRAVIR, the research activity "Active Vision Sensor" is currently developed by François Berry (MCF, LASMEA). However, other works dealing with the problem of front vehicle tracking have defined a specific active vision sensor coupling a PTZ camera with a wide focal length classical camera.

Abstract
Active Vision Sensor.
  • Front Vehicle Tracking : a specific vision sensor coupling a PTZ (Pan-Tilt-Zoom) camera with a wide focal length classical camera has been defined.

    A wide angle camera coupled with a PTZ camera


    The wide angle camera perfoms a global perception task, in order to :
    • define the area where the pattern to be track is present (end of road, front car vehicle)
    • inialize the PTZ parameters in order to perform tracking (front car vehicle)
    • perform current car and road localization (in order to elect the most dangerous front car vehicle)
    • reactualize the target to be tracked depending on traffic evolution
    The PTZ camera is programmed in order to focalize (same size and position in image) the current image on a predefined image.

    Synoptic of global software environment



    Applications : End of road tracking - Front car tracking


    The main goals of such system is to deliver safety informations in real time to the driver (driving assistance system), or in the future to adapt longitudinal speed of the current vehicle regarding the front car vehicle (Adaptive Cruise Control System). The lateral deviation can also be controled by using the wide angle camera.

    (Video1, Video2)
    Left view : wide angle camera, Right view : PTZ camera
    Front truck tracking and camping car tracking


    The tracking algorithm has been studied by F. Jurie and M. Dhome in LASMEA. It has been used and adapted to the particular case of front car tracking.

    Synoptic of the tracking algorithm


    The real demonstrator VELAC is a 806 from Peugeot SA.

    LASMEA real demonstrator : VELAC


  • Design of a novel Active Vision Sensor using CMOS technology : a SOC architecture integrates a FPGA Stratix from Altera, and different CMOS retinas.

    The Active Vision Sensor is composed of :
    • a mother board where the STRATIX FPGA is integrated
    • a daughter board where the CMOS retina is placed

    Specific developments are done using an Architecture Algorithm Adequation approach. The Active Vision Sensor can be connected to a host PC computer through Parallel Bus, USB2 and IEEE1394A/B (soon).
    For more details or advanced informations about this development, please contact F. Berry in LASMEA.

    Photo of the novel Active Vision Sensor


Project and Contract
Active Vision Sensor.
  • CPER-MSPI (01-03) project : Kinematic and dynamic identification of High Speed Machines : using high speed vision metrology
People
Active Vision Sensor The list of participating people is :
  • LASMEA
    • Researchers
      F. Berry, P. Martinet,
    • PhD
      P. Chalimbeau,
  • Former collaborators
    • Researcher F. Collange, F. Jurie, J. Alizon,
    • PhD X. Clady (03),
    • Master P. Chalimbeau (01),
Publications
  • Phd thesis
    • Pierre Chalimbeau, “Conception d'un capteur pour la vision active”, LASMEA, Blaise Pascal University, Clermont-Ferrand, to be done in 2004
    • Xavier Clady, “Contributions à la navigation autonome d'un véhicule automobile par vision”, LASMEA, Blaise Pascal University, Clermont-Ferrand, January 6th, 2003
  • Journal
    • P. Chalimbaud , F. Berry, “Smart Camera and Active Vision: The active detector formalism”, to appear in SPIE Newsletters on Electronic Imaging, 2004
  • Book, Book Chapter, Tutorials, Report
    • none
  • Conferences
    • P. Chalimbaud , F. Berry, “Conception d’un capteur de vision active basé sur un imageur CMOS”, in Proceedings of the RFIA conference, RFIA'04, pp., Toulouse, France, January 28-30th, 2004
    • X. Clady, F. Collange, F. Jurie, P. Martinet, “Cars detection and tracking with a vision sensor”, in Proceedings of the IEEE Intelligent Vehicles Symposium, IV'03, pp. 593-598, Colombus, Ohaio, USA, June 9-11th, 2003
    • P. Chalimbaud , F. Berry, P. Martinet, “The task "template tracking" in a sensor dedicated to active vision”, in Proceedings of the Proceedings of the Workshop on Computer Architecture for Machine Perception , CAMP'03, pp. , News Orleans, Louisiana, USA, 2003
    • P. Chalimbaud , F. Berry , P. Martinet, “Active vision based on CMOS camera : an active detector approach”, in Proceedings of the International Symposium on Robots, ISR'02, CDROM , Stockholm, Sweden, October 7-11th, 2002
    • T. Chateau, F. Jurie, M. Dhome, X. Clady, “Real-time tracking using wavelet representation”, in Proceedings of the Deutsche Arbeitsgemeinshaft fur Mustererkennung Symposium, DAGM'02, pp. 523-530, Zurich, Switzermland, September, 2002
    • X. Clady, F. Collange, F. Jurie, B. Thuilot, P. Martinet, “Détection et suivi de véhicules par vision”, in Proceedings of the Colloque GRETSI sur le Traitement du Signal et de l'Image, GRETSI'01, CD-ROM , Toulouse, France, September 10-13th, 2001
    • X. Clady, F. Collange, F. Jurie, P. Martinet, “Tracking with a pan-tilt-zoom camera for an ACC system”, in Proceedings of the 12th IAPR Scandinavian Conference on Computer Vision, ISC-CV'01, pp. 561-566, Bergen, Norway, June, 2001
    • X. Clady, F. Collange, F. Jurie, P. Martinet, “Object Tracking with a Pan Tilt Zoom Camera application to car driving assistance”, in Proceedings of the IEEE International Conference on Robotics and Automation, ICRA'01, vol. 2, pp. 1653 -1658, Seoul, Korea, May 23-25th, 2001