ECE PhD Prospectus Defense Announcement: Joshua Rapp, 12/20, 10:00 AM, PHO 404/428
- Starts: 10:00 am on Wednesday, December 20, 2017
- Ends: 12:00 pm on Wednesday, December 20, 2017
Title: Computational Methods for Photon-Efficient Lidar Imaging Presenter: Joshua Rapp Date: Wednesday, December 20, 2017 Time: 10 AM - 12 PM Location: 8 St. Mary’s Street, PHO 404/428 Advisor/Chair: Vivek Goyal (ECE) Committee: Vivek Goyal (ECE), Robin Dawson (Draper Labs), Clem Karl (ECE), Lei Tian (ECE) Abstract: Lidar is an increasingly prevalent technology for depth sensing, with applications including scientific measurement, covert imaging, and autonomous navigation systems. Recent results using single-photon avalanche diode (SPAD) detectors have shown lidar depth and reflectivity images formed from as little as one photon detection per pixel by precisely modeling the probability of photon detections and exploiting the structure of natural scenes. Reducing the number of photons needed for accurate image formation enables faster, farther, and safer acquisition. Still, such photon-efficient systems are limited to laboratory conditions more favorable than the real-world settings in which they would be deployed. We propose a suite of methods that jointly modify the hardware and signal processing to surpass the performance of state-of-the-art photon counting systems. For instance, we achieve robustness to high levels of ambient light by introducing a window-based censoring method to separate signal and background light detections. Spatial correlations both within and between depth and reflectivity images are encoded in superpixel constructions, which fill in holes caused by the censoring. Accurate depth and reflectivity images can then be formed with an average of 2 signal photons and 50 background photons per pixel. To improve temporal resolution especially for SPAD arrays, we propose a subtractively-dithered lidar implementation, which uses changing synchronization delays to shift the time-quantization bin edges. We examine the noise model resulting from dithering when the laser has a Gaussian pulse and introduce depth estimators that take advantage of this model. Finally, we consider solutions to further problems, such as extending the lidar system’s unambiguous range, correcting for SPAD dead time effects, and interpreting indirect light reflections from objects around corners.
- 8 Saint Mary's St. Boston, MA Room 404/428