Passive Multidimensional Imaging and Recognition with Multiple Degrees of Freedom
Abstract
The purpose of this project is to introduce, investigate, and demonstrate a new multi-dimensional sensing capability where a small number of photons may allow the possibility of implementing reliable multidimensional visualization, identification, and classification using polarimetric and multi-spectral band passive imaging. The multidimensional sensing and imaging proposed here are intended to operate under extremely sparse (low photon count) data conditions to enable many benefits, including operation in degraded environments, removing of occlusion to allow visualization of occluded objects, multi-dimensional target recognition, and tracking of occluded objects. Unlike LADAR, this passive multi-dimensional sensing approach uses 2D image sensors with ambient light or thermal imaging for persistent and covert operation. This capability will be very beneficial to enhancing surveillance, monitoring human activities, and information exploitation. During the course of this project, we have performed experiments and theoretical analysis of the objectives and tasks of the project. We have investigated polarimetric multidimensional imaging in degraded environments. We have investigated deep convolutional neural network model using physics based training with three-dimensional (3D) imaging to recognize polarimetric 3D objects in degraded environments such as low light and partial occlusions.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 18, 2022
- Accession Number
- AD1190034
Entities
People
- Bahram Javidi
Organizations
- University of Connecticut