טכניון מכון טכנולוגי לישראל
הטכניון מכון טכנולוגי לישראל - בית הספר ללימודי מוסמכים  
M.Sc Thesis
M.Sc StudentMonin Sagi
SubjectRapid Single-Pixel Imaging
DepartmentDepartment of Electrical Engineering
Supervisor Professor Amir Rosenthal
Full Thesis textFull thesis text - English Version


Abstract

Single pixel imaging (SPI) is an optical imaging technique that uses a single detector instead of a digital camera, which contains an array of detectors. The image in SPI is obtained by scanning the object with different spatial pattern and measuring the intensity collected with the single detector. This imaging technique has attracted much interest in the last decade with many applications demonstrated. Despite its potential, SPI still suffers from several drawbacks limiting its application in real-world imaging applications. One of the major limitations of SPI is its long image acquisition time, determined by the refresh rate of spatial light modulators, conventionally limited to 22 kHz. As a result, SPI cannot be used for imaging dynamic objects without significantly compromising on resolution or image quality.

This research thesis is dedicated to the development of experimental and algorithmic techniques to enable SPI of dynamic objects. In our experimental work, we develop a new approach to produce the spatially modulated patterns, leading to pattern switching at rates of up to 2.4 MHz - 2 orders of magnitude faster than conventional technology. The fast acquisition of our system is the key to our algorithmic work. With fast acquisition the motion of the scene is relatively small between subsequent frames. In our algorithmic work, we incorporate motion estimation and correction to minimize artifacts and blur effects when imaging moving objects. Our algorithms are based on analyzing variations in subsequent frames of SPI videos, they can be used to counter both global and local motion. We successfully demonstrate our algorithms on both numerically simulations and experimental data.