|M.Sc Student||Shoham Avi|
|Subject||Fast Optical CMOS Sensor-Based Applications and|
Signal Preprocessing Algorithms
|Department||Department of Electrical Engineering||Supervisor||Professor Emeritus Yael Nemirovsky|
|Full Thesis text|
CMOS Image Sensors and Single-Photon Avalanche Diodes are two state-of-the-art technologies for detection of photons in the visual and near-infrared wavelengths. The two technologies differ in their sensitivity to photons and in the properties of their pixels but could both be used for the design of high speed sensors and sensors for photon-starved scenarios.
This work focuses on the usage of these two types of sensors for detection of gun muzzle flashes. The work establishes a physical model for the prediction of performance of systems designed, based on these sensors, for the detection of muzzle flashes based on the emission of photons from excited Alkali salts located in the ammunition gunpowder. The influence of different system parameters (sampling frequency, optical aperture, effective focal length, field-of-view, spectral width), as well as scenario parameters (illumination level, atmospheric attenuation and ambient temperature) are discussed. The model is then validated based on radiometric measurements.
Several signal pre-processing algorithms for the detection of gun muzzle flashes are described and their performance, in terms of false alarm rates (on background recordings) and detection probability (of true events), is analyzed and discussed. A proof-of-concept, in which one of the algorithms was implemented and used in real-time scenarios is described.