Abstract: A method for designing a spectral sensing device. The method includes: (1) performing computational operations on a computer, wherein the computational operations determine the positions of diffracted orders of an optical system model that models at least an array of light modulating elements and a diffraction grating, wherein the diffracted orders correspond to respective spectral components of input light to the optical system model, wherein the positions of the diffracted orders are determined at a target plane of the optical system model; and (2) storing the positions of the diffracted orders in a memory, wherein the positions determine corresponding locations for light detectors in the spectral sensing device. The spectral sensing device may be assembled by modifying an existing single pixel camera, i.e., by adding the diffraction grating and adding the light detectors respectively at said positions of the diffracted orders.

Abstract: A mechanism for reconstructing a signal (e.g., an image) based on a vector s, which includes measurements of the signal. The measurements have been acquired using at least a portion of a measurement vector set represented by a matrix H. Each of the measurements corresponds to a respective row of the matrix H. (For example, each of the measurements may correspond to an inner product between the signal and a respective row of the matrix product HD, wherein D is a generalized permutation matrix.) A total-variation primal-dual hybrid gradient (TV-PDHG) algorithm is executed based on data including the matrix H and the vector s, to determine an estimate for the signal. The TV-PDHG algorithm is implemented in fixed-point arithmetic.

Abstract: A compressive imaging system and method for quickly detecting spectrally and spatially localized events (such as explosions or gun discharges) occurring within the field of view. An incident light stream is modulated with a temporal sequence of spatial patterns. The wavelength components in the modulated light stream are spatially separated, e.g., using a diffractive element. An array of photodetectors is used to convert subsets of the wavelength components into respective signals. An image representing the field of view may be reconstructed based on samples from some or all the signals. A selected subset of the signals are monitored to detect event occurrences, e.g., by detecting sudden changes in intensity. When the event is detected, sample data from the selected subset of signals may be analyzed to determine the event location within the field of view. The event location may be highlighted in an image being generated by the imaging system.

Abstract: A methodology for acquiring measurements of a signal at one or more scales of resolution, including: generating modulation patterns based on a predefined measurement matrix; modulating a received signal with the modulation patterns using the signal modulating array to obtain a modulated signal; and acquiring measurements of intensity of the modulated signal. Each modulation pattern is generated by: (a) selecting a corresponding row of the measurement matrix; (b) reordering elements of the selected row according to a permutation to obtain a reordered row; and (c) transferring the reordered row to the signal modulating array so that elements of the reordered row are mapped onto the signal modulating array. The permutation is defined so that a subset of the modulation patterns are coarse patterns that respect a partition of the signal modulating array into an array of superpixels, each superpixel including a respective group of the signal modulating elements.

Abstract: A compressive imaging system modulates an incident light stream and senses the modulated light stream to obtain compressive measurements. The measurements are algorithmically processed to reconstruct a sequence of images. The image sequence is displayed. The system receives user input (through a user interface) representing a user command to set or change one or more visual performance factors such as image quality and frame rate. The system immediately adjusts the visual performance factors by adjusting one or more underlying system parameters/algorithms. Thus, the visual consequences of any inputs to the user interface become immediately apparent in the displayed sequence of images. The user may therefore intuitively learn how to operate the user interface simply by making trial inputs and observing their effects in the displayed image sequence. The user interface may include one or more mechanical input devices and/or one or more graphical user interface (GUI) elements.

Abstract: If a Hadamard matrix HN of order N=BF is a Kronecker product HFHB of an order F Hadamard matrix and an order B Hadamard matrix, then transformation by HN may be implemented by a fast Hadamard transform at coarse scale followed by fast Hadamard transforms at fine scale. Alternatively, transformation by HN may be achieved by performing order B transforms on columns of a two-dimensional array and order B transforms on rows of the array. As another alternative, transformation by HN may be achieved by computing intermediate values based on linear combinations of input elements and then computing linear combinations of the intermediate values. For compressive signal acquisition, any row of HN may be generated by concatenating selectively modified copies of a corresponding row of HB. Thus, modulation patterns may be generated on the fly.

Abstract: A technique to collect measurements that are adapted to a signal/scene of interest is presented. The measurements are correlations with patterns that serve as modulating waveforms. The patterns correspond respectively to rows of a sensing matrix. The method uses a sensing matrix whose rows are partitioned into blocks. Each block corresponds to a distinct feature or salient property of the scene. For each block, the method collects a number of measurements of the signal/scene based on selected rows of the block, and generates one or more associated statistics for the block based on said measurements. The statistics for the blocks are then analyzed (e.g., sorted) to determine the most important blocks. Subsequent measurements of the signal/scene may be based on rows from those most important blocks. The original measurements and/or the subsequent measurements may then be used in an algorithm to reconstruct the signal/scene.

Abstract: A compressive imaging system for optimizing dynamic range during the acquisition of compressed images. A light modulator modulates incident light with spatial patterns to produced modulated light. A light sensing device generates an electrical signal representing intensity of the modulated light over time. The system amplifies a difference between the electrical signal and an adjustable baseline voltage and captures samples of the amplified signal. The adjustable baseline voltage is set to be approximately equal to the average value of the electrical signal. A compressive imaging system for identifying and correcting hot spot(s) in the incident light field. Search patterns are sent to the light modulator and the corresponding samples of the electrical signal are analyzed. Once the hot spot is located, the light modulating elements corresponding to the hot spot may be turned off or their duty cycle may be reduced.

Abstract: An imaging system and method that captures compressive sensing (CS) measurements of a received light stream, and also obtains samples of background light level (BGLL). The BGLL samples may be used to compensate the CS measurements for variations in the BGLL. The system includes: a light modulator to spatially modulate the received light stream with spatial patterns, and a lens to concentrate the modulated light stream onto a light detector. The samples of BGLL may be obtained in various ways: (a) injecting calibration patterns among the spatial patterns; (b) measuring complementary light reflected by digital micromirrors onto a secondary output path; (c) separating and measuring a portion of light from the optical input path; (d) low-pass filtering the CS measurements; and (e) employing a light power meter with its own separate input path. Also, the CS measurements may be high-pass filtered to attenuate background light variation.

Abstract: Mechanisms for increasing the rate of acquisition of compressed/encoded image representations are disclosed. An imaging system may deliver subsets of a modulated light stream onto respective light sensing devices. The light sensing devices may be sampled in parallel. Samples from each light sensing device may be used to construct a respective sub-image of a final image. The parallelism allows compressed images to be acquired at a higher rate. The number of light sensing devices and/or the number of pixels per image may be selected to achieve a target image acquisition rate. In another embodiment, spatial portions of the incident light stream are separated and delivered to separate light modulators. In yet another embodiment, the incident light stream is split into a plurality of beams, each of which retains the image present in the incident light stream and is delivered to a separate light modulator.

Abstract: A mechanism for reconstructing a signal (e.g., an image) based on a vector s, which includes measurements of the signal. The measurements have been acquired using at least a portion of a measurement vector set represented by a matrix H. Each of the measurements corresponds to a respective row of the matrix H. (For example, each of the measurements may correspond to an inner product between the signal and a respective row of the matrix product HD, wherein D is a generalized permutation matrix.) A total-variation primal-dual hybrid gradient (TV-PDHG) algorithm is executed based on data including the matrix H and the vector s, to determine an estimate for the signal. The TV-PDHG algorithm is implemented in fixed-point arithmetic.

Abstract: If a Hadamard matrix HN of order N=BF is a Kronecker product HF HB of an order F Hadamard matrix and an order B Hadamard matrix, then transformation by HN may be implemented by a fast Hadamard transform at coarse scale followed by fast Hadamard transforms at fine scale. Alternatively, transformation by HN may be achieved by performing order B transforms on columns of a two-dimensional array and order B transforms on rows of the array. As another alternative, transformation by HN may be achieved by computing intermediate values based on linear combinations of input elements and then computing linear combinations of the intermediate values. For compressive signal acquisition, any row of HN may be generated by concatenating selectively modified copies of a corresponding row of HB. Thus, modulation patterns may be generated on the fly.

Abstract: A mechanism for efficiently loading rows of an N×N transform matrix HN into a signal-modulating array. A row index m(i) that identifies a row r[m(i)] of HN is generated, where i is in the range {0, 1, . . . , L?1}; L is less than or equal to B; and m(i) is in the range {0, 1, . . . , B?1}. HN has the form HN=HFHB. HF is an F×F matrix, and HB is a B×B matrix. denotes the Kronecker product. The row r[m(i)] of HN is generated and loaded into the array. For each k in the range {1, 2, . . . , F?1}, a row r[m(i)+kB] from HN is partially loaded into the array by: loading a first subset of elements of row r[m(i)+kB] that are not currently present in the array, and not loading a second subset of elements of row r[m(i)+kB] that are currently present in the array.

Abstract: A methodology for acquiring measurements of a signal at one or more scales of resolution, including: generating modulation patterns based on a predefined measurement matrix; modulating a received signal with the modulation patterns using the signal modulating array to obtain a modulated signal; and acquiring measurements of intensity of the modulated signal. Each modulation pattern is generated by: (a) selecting a corresponding row of the measurement matrix; (b) reordering elements of the selected row according to a permutation to obtain a reordered row; and (c) transferring the reordered row to the signal modulating array so that elements of the reordered row are mapped onto the signal modulating array. The permutation is defined so that a subset of the modulation patterns are coarse patterns that respect a partition of the signal modulating array into an array of superpixels, each superpixel including a respective group of the signal modulating elements.

Abstract: An imaging system and method that captures compressive sensing (CS) measurements of a received light stream, and also obtains samples of background light level (BGLL). The BGLL samples may be used to compensate the CS measurements for variations in the BGLL. The system includes: a light modulator to spatially modulate the received light stream with spatial patterns, and a lens to concentrate the modulated light stream onto a light detector. The samples of BGLL may be obtained in various ways: (a) injecting calibration patterns among the spatial patterns; (b) measuring complementary light reflected by digital micromirrors onto a secondary output path; (c) separating and measuring a portion of light from the optical input path; (d) low-pass filtering the CS measurements; and (e) employing a light power meter with its own separate input path. Also, the CS measurements may be high-pass filtered to attenuate background light variation.

Abstract: A mechanism for reconstructing sub-images based on measurement data acquired by an imaging system including an array of light modulating elements and an array of photodetectors. Each sub-image is reconstructed based on samples from a respective photodetector and a respective set of measurement patterns defined on a respective virtual sub-region on the modulating array. Each virtual sub-region is configured to include at least the light modulating elements that are able to send a non-trivial amount of light to the respective photodetector during a pattern application period. The virtual sub-regions overlap because many light modulating elements are capable of sending light to more than one photodetector. Whenever a measurement pattern of one virtual sub-region overlaps the measurement pattern of a neighboring virtual sub-region, the two measurement patterns agree by design. Thus, the measurement patterns for the collection of virtual sub-regions combine to form a pattern on the whole modulating array.

Abstract: A system and method for searching an incident light field for atypical regions (e.g., hot spots or cool spots or spectrally distinctive regions) within the incident light field using a light modulator and a spectral sensing device. Once the atypical regions are identified, the light modulator may be used to mask the incident light field so that the spectral sensing device can make spatially-concentrated measurements of the wavelength spectrum of the atypical regions (or alternatively, the exterior of the atypical regions). Furthermore, in a compressive imaging mode, a sequence of spatial patterns may be supplied to the light modulator, and a corresponding sequence of wavelength spectra may be collected from the spectral sensing device. The wavelength spectra comprise a compressed representation of the incident light field over space and wavelength. The wavelength spectra may be used to reconstruct a multispectral (or hyperspectral) data cube.

Abstract: An imaging system and method that captures compressive sensing (CS) measurements of a received light stream, and also obtains samples of background light level (BGLL). The BGLL samples may be used to compensate the CS measurements for variations in the BGLL. The system includes: a light modulator to spatially modulate the received light stream with spatial patterns, and a lens to concentrate the modulated light stream onto a light detector. The samples of BGLL may be obtained in various ways: (a) injecting calibration patterns among the spatial patterns; (b) measuring complementary light reflected by digital micromirrors onto a secondary output path; (c) separating and measuring a portion of light from the optical input path; (d) low-pass filtering the CS measurements; and (e) employing a light power meter with its own separate input path. Also, the CS measurements may be high-pass filtered to attenuate background light variation.

Abstract: A compressive imaging system including a light modulator, a light sensing device and a TIR prism. The TIR prism is configured to receive an incident light beam, to provide the incident light beam to the light modulator, to receive a modulated light beam MLB from the light modulator, and to direct the modulated light beam onto a sensing path. The light sensing device receives the modulated light beam (or at least a portion of the modulated light beam) and generates an electrical signal that represents intensity of the modulated light beam (or the “at least a portion” of the modulated light beam). The TIR prism may reduce a distance required to separate the incident light beam from the modulated light beam.

Abstract: A compressive imaging (CI) device for attenuating noise. The CI device may acquire samples during steady state portions of pattern modulation periods, avoiding the disturbing effect of transients that occur at pattern transitions. A CI device may acquire and then average multiple samples per spatial pattern to reduce (deterministic and/or random) zero-mean noise. A CI device may apply a filter to the photodetector signal in the analog domain and/or in the digital domain to attenuate noise components, e.g., noise due to electromagnetic interference.