Single-Pixel Imaging
An imaging technique in which a scene is encoded and measured through a series of binary illumination patterns and an image is then obtained by computational reconstruction. The design of adaptive measurements in this setting faces combinatorial-explosion computational challenges.
Timeline (last 90 days)
Compressed sensing is one of the theoretical foundations of single-pixel imaging; it points out that if a signal is sparse in some transform domain, it can be recovered with far fewer measurements than required by the Nyquist sampling theorem.
LIPS-PC (Learning Illumination Patterns for Single-Pixel Classification) is a single-pixel imaging method that directly learns illumination patterns for classification, without the need to first reconstruct the image and then classify it.
Single-pixel imaging (ghost imaging) uses a spatial light modulator to project known patterns and a single-point detector to measure the total light intensity, reconstructing the image by solving a linear system of equations or an optimization problem.
Single-pixel imaging offers significant cost advantages in non-visible bands (such as X-ray, terahertz, and short-wave infrared), because high-resolution array detectors in these bands are often expensive or even unavailable.
Single-pixel imaging (SPI) encodes a scene through a series of binary illumination patterns and then obtains an image via computational reconstruction.
A recent work reformulates the Bayesian sequential optimal experimental design (sOED) problem as a reinforcement learning (RL) task for adaptive measurement in single-pixel imaging.
The design space of single-pixel imaging consists of all possible binary illumination patterns; if an image has N pixels, the number of possible patterns is 2^N; for a 32×32 image (N=1024), the design space reaches a scale of 2^1024, far beyond the enumerable range.
The study adopts a simple but representative single-pixel imaging scenario and evaluates the scalability of the learned policy under two different policy parameterizations.
All Facts (8)
Compressed sensing is one of the theoretical foundations of single-pixel imaging; it points out that if a signal is sparse in some transform domain, it can be recovered with far fewer measurements than required by the Nyquist sampling theorem.
50%UnverifiedLIPS-PC (Learning Illumination Patterns for Single-Pixel Classification) is a single-pixel imaging method that directly learns illumination patterns for classification, without the need to first reconstruct the image and then classify it.
50%UnverifiedSingle-pixel imaging (ghost imaging) uses a spatial light modulator to project known patterns and a single-point detector to measure the total light intensity, reconstructing the image by solving a linear system of equations or an optimization problem.
50%UnverifiedSingle-pixel imaging offers significant cost advantages in non-visible bands (such as X-ray, terahertz, and short-wave infrared), because high-resolution array detectors in these bands are often expensive or even unavailable.
50%UnverifiedSingle-pixel imaging (SPI) encodes a scene through a series of binary illumination patterns and then obtains an image via computational reconstruction.
50%UnverifiedA recent work reformulates the Bayesian sequential optimal experimental design (sOED) problem as a reinforcement learning (RL) task for adaptive measurement in single-pixel imaging.
50%UnverifiedThe design space of single-pixel imaging consists of all possible binary illumination patterns; if an image has N pixels, the number of possible patterns is 2^N; for a 32×32 image (N=1024), the design space reaches a scale of 2^1024, far beyond the enumerable range.
50%UnverifiedThe study adopts a simple but representative single-pixel imaging scenario and evaluates the scalability of the learned policy under two different policy parameterizations.
50%