In the past decades, time-resolved imaging such as fluorescence lifetime or time-of-flight depth imaging has been extensively explored in biomedical and industrial fields because of its non-invasive characterization of material properties and remote sensing capability. Many studies have shown its potential and effectiveness in applications such as cancer detection and tissue diagnoses from fluorescence lifetime imaging, and gesture/motion sensing and geometry sensing from time-of-flight imaging. Nonetheless, time-resolved imaging has not been widely adopted due to the high cost of the system and performance limits.The research presented in this thesis focuses on the implementation of low-cost real-time time-resolved imaging systems. Two image sensing schemes are proposed and implemented to address the major limitations. First, we propose a single-shot fluorescence lifetime image sensors for high speed and high accuracy imaging. To achieve high accuracy, previous approaches repeat the measurement for multiple sampling, resulting in long measurement time. On the other hand, the proposed method achieves both high speed and accuracy at the same time by employing a pixel-level processor that takes and compresses the multiple samples within a single measurement time. The pixels in the sensor take multiple samples from the fluorescent optical signal in sub-nanosecond resolution and compute the average photon arrival time of the optical signal. Thanks to the multiple sampling of the signal, the measurement is insensitive to the shape or the pulse-width of excitation, providing better accuracy and pixel uniformity than conventional rapid lifetime determination (RLD) methods. The proposed single-shot image sensor also improves the imaging speed by orders of magnitude compared to other conventional center-of-mass methods (CMM).Second, we propose a 3-D camera with a background light suppression scheme which is adaptable to various lighting conditions. Previous 3-D cameras are not operable in outdoor conditions because they suffer from measurement errors and saturation problems under high background light illumination. We propose a reconfigurable architecture with column-parallel discrete-time background light cancellation circuit. Implementing the processor at the column level allows an order of magnitude reduction in pixel size as compared to existing pixel-level processors. The column-level approach also provides reconfigurable operation modes for optimal performance in all lighting conditions. For example, the sensor can operate at the best frame-rate and resolution without the presence of background light. If the background light saturates the sensor or increases the shot noise, the sensor can adjust the resolution and frame-rate by pixel binning and superresolution techniques. This effectively enhances the well capacity of the pixel to compensate for the increase shot noise, and speeds up the frame processing to handle the excessive background light. A fabricated prototype sensor can suppress the background light more than 100-klx while achieving a very small pixel size of 5.9μm.
PDF
download
878987797
028-85220240
