Countries are trying to equip their public transportation infrastructure with fixed radiation portals and detectors to detect radiological threat. Current works usually focus on neutron detection, which could be useless in the case of dirty bomb that would not use fissile material. Another approach, such as gamma dose rate variation monitoring is a good indication of the presence of radionuclide. However, some legitimate products emit large quantities of natural gamma rays; environment also emits gamma rays naturally. They can lead to false detections. Moreover, such radio-activity could be used to hide a threat such as material to make a dirty bomb. Consequently, radionuclide identification is a requirement and is traditionally performed by gamma spectrometry using unique spectral signature of each radionuclide. These approaches require high-resolution detectors, sufficient integration time to get enough statistics and large computing capacities for data analysis. High-resolution detectors are fragile and costly, making them bad candidates for large scale homeland security applications. Plastic scintillator and NaI detectors fit with such applications but their resolution makes identification difficult, especially radionuclides mixes. This paper proposes an original signal processing strategy based on artificial spiking neural networks to enable fast radionuclide identification at low count rate and for mixture. It presents results obtained for different challenging mixtures of radionuclides using a NaI scintillator. Results show that a correct identification is performed with less than hundred counts and no false identification is reported, enabling quick identification of a moving threat in a public transportation. Further work will focus on using plastic scintillators.
Today's digital image sensors are used as passive photon integrators and image processing is essentially performed
by digital processors separated from the image sensing part. This approach imposes to the processing part to
deal with already grabbed pictures with possible unadjusted exposition parameters. This paper presents a fast
self-adaptable preprocessing architecture with fast feedbacks on the sensing level. These feedbacks are controlled
by digital processing in order to modify the sensor parameters during exposure time. Exposition and processing
parameters are tuned in real life to fit with applications requirement depending on scene parameters. Considering
emerging integration technologies such as 3D stacking, this paper presents an innovative way of designing smart
vision sensors, integrating feedback control and opening new approaches for machine vision architectures.
The advent of camera phones marks a new phase in embedded camera sales. By late 2009, the total number of camera
phones will exceed that of both conventional and digital cameras shipped since the invention of photography. Use in mobile
phones of applications like visiophony, matrix code readers and biometrics requires a high degree of component flexibility
that image processors (IPs) have not, to date, been able to provide. For all these reasons, programmable processor solutions
have become essential. This paper presents several techniques geared to speeding up image processors. It demonstrates
that a gain of twice is possible for the complete image acquisition chain and the enhancement pipeline downstream of the
video sensor. Such results confirm the potential of these computing systems for supporting future applications.