Method for acquiring intentionally limited data and the machine learning approach to reconstruct it
AssigneeNorthwestern University
InventorsKarl A. Hujsak, Vinayak P. Dravid, Benjamin D. Myers
Life of this patent
Estimating the term…
Abstract
Aspects of the present disclosure involve a data capturing and processing system that intentionally captures data and/or data sets with missing pieces of information. The data and/or datasets may include various types of data, such as one-dimensional signals, two-dimensional images (or other images), and/or three-dimensional structures. The captured data is processed to restore missing information into the data and/or data sets, thereby enabling simultaneous pattern recognition and image recovery.
Claims16
A system for reconstructing images comprising:
- a measurement system configured to capture at least one image corresponding to a sample specimen, wherein the measurement system includes a high-speed electrostatic beam blanker that includes a deflector that blocks an electron beam, which prevents portions of the sample specimen from being imaged such that portions of the at least one image are missing due to intentional under-sampling of the sample specimen, and wherein the measurement system is configured to control the high-speed electrostatic beam blanker to intentionally under-sample the sample specimen based on a pre-defined percentage-of-pixel value and a pre-defined dwell time;
- and
- a processing device, in operable communication with the measurement system, to automatically restore the missing portions of the at least one image, wherein the processing device is configured to:
- access a learned dictionary to identify latent features of the at least one image;
- identify linear combinations of the latent features identified from the learned dictionary;
- and
- reconstruct the missing portions of the at least one image according to the one or more linear combinations.
A method for reconstructing images comprising:
- capturing, using a measurement system, at least one image corresponding to a sample specimen, wherein the capturing includes using a high-speed electrostatic beam blanker that includes a deflector to block an electrostatic beam, which prevents portions of the sample specimen from being imaged such that portions of the at least one image are missing due to intentional under-sampling of the sample specimen, and wherein the capturing includes controlling the high-speed electrostatic beam blanker to intentionally under-sample the sample specimen based on a pre-defined percentage-of-pixel value and a pre-defined dwell time;
- and
- automatically restoring, using a processing device, the missing portions of the at least one image by:
- accessing a learned dictionary to identify latent features of the at least one image;
- identifying linear combinations of the latent features that were identified from the learned dictionary;
- and
- reconstructing the missing portions of the at least one image according to the one or more linear combinations.
A non-transitory computer readable medium encoded with instructions for reconstructing images, the instructions, executable by a processing device, comprising:
- capturing, at a measurement system, at least one image corresponding to a sample specimen, wherein the capturing includes using a high-speed electrostatic beam blanker that includes a deflector to block an electrostatic beam, which prevents portions of the sample specimen from being imaged such that portions of the at least one image are missing due to intentional under-sampling of the sample specimen, and wherein the capturing includes controlling the high-speed electrostatic beam blanker to intentionally under-sample the sample specimen based on a pre-defined percentage-of-pixel value and a pre-defined dwell time;
- and
- automatically restoring, using the processing device, the missing portions of the at least one image by:
- accessing a learned dictionary to identify latent features of the at least one image;
- identifying linear combinations of the latent features identified from the learned dictionary;
- and
- reconstructing the missing portions of the at least one image according to the one or more linear combinations.
Dependencies are inferred from the claim text.