Photon-efficient super-resolution laser radar
The resolution achieved in photon-efficient active optical range imaging systems can be low due to non-idealities such as propagation through a diffuse scattering medium. We propose a constrained optimization-based frame- work to address extremes in scarcity of photons and blurring by a forward imag...
Main Authors: | , , |
---|---|
Other Authors: | , |
Format: | Article |
Language: | English |
Published: |
SPIE-Intl Soc Optical Eng,
2018-06-25T18:11:16Z.
|
Subjects: | |
Online Access: | Get fulltext |
Summary: | The resolution achieved in photon-efficient active optical range imaging systems can be low due to non-idealities such as propagation through a diffuse scattering medium. We propose a constrained optimization-based frame- work to address extremes in scarcity of photons and blurring by a forward imaging kernel. We provide two algorithms for the resulting inverse problem: a greedy algorithm, inspired by sparse pursuit algorithms; and a convex optimization heuristic that incorporates image total variation regularization. We demonstrate that our framework outperforms existing deconvolution imaging techniques in terms of peak signal-to-noise ratio. Since our proposed method is able to super-resolve depth features using small numbers of photon counts, it can be useful for observing fine-scale phenomena in remote sensing through a scattering medium and through-the-skin biomedical imaging applications. Samsung Scholarship Foundation National Science Foundation (U.S.) (Grant 1161413) National Science Foundation (U.S.) (Grant 1422034) |
---|