Hyperspectral imaging devices using hybrid vector and tensor processing
View Patent ↗Methods and systems obtain data representative of a scene across spectral bands using a compressive-sensing-based hyperspectral imaging system comprising optical elements. These methods and systems sample two modes of a three-dimensional tensor corresponding to a hyperspectral representation of the scene using sampling matrices, one for each of the two modes, to generate a modified three-dimensional tensor. After sampling the two modes, such methods and systems sample a third mode of the modified three-dimensional tensor using a third sampling matrix to generate a further modified three-dimensional tensor. Then, the methods and systems reconstruct hyperspectral data from the further modified three-dimensional tensor using the sampling matrices and the third sampling matrix.
1. A method comprising:
obtaining data representative of a scene across spectral bands using a compressive-sensing-based hyperspectral imaging system comprising optical elements;
sampling two modes of a three-dimensional tensor corresponding to a hyperspectral representation of said scene using sampling matrices, one for each of said two modes, to generate a modified three-dimensional tensor;
after said sampling two modes, sampling a third mode of said modified three-dimensional tensor using a third sampling matrix to generate a further modified three-dimensional tensor; and
reconstructing hyperspectral data from said further modified three-dimensional tensor using said sampling matrices and said third sampling matrix,
said reconstructing comprising recovering each mode of said three-dimensional tensor via a sequence of optimization processing in a reverse order of said sampling two modes and said sampling a third mode, said sequence comprising:
applying an optimization process to said further modified three-dimensional tensor to reconstruct said modified three-dimensional tensor as a recovered modified three-dimensional tensor; and
applying an optimization process to said recovered modified three-dimensional tensor to reconstruct said three-dimensional tensor corresponding to said data representative of said scene.
2. The method according to claim 1 , said applying an optimization process to said further modified three-dimensional tensor comprising performing vectorial optimization on said third mode of said further modified three-dimensional tensor, and
said applying an optimization process to said recovered modified three-dimensional tensor comprising performing joint tensorial optimization on said two modes of said recovered modified three-dimensional tensor.
3. The method according to claim 2 , said joint tensorial optimization being one of parallelizable joint tensorial optimization and serial joint tensorial optimization.
4. A method comprising:
obtaining data representative of a scene across spectral bands using a compressive-sensing-based hyperspectral imaging system comprising optical elements;
sampling two modes of a three-dimensional tensor corresponding to a hyperspectral representation of said scene using sampling matrices, one for each of said two modes, to generate a modified three-dimensional tensor;
after said sampling two modes, sampling a third mode of said modified three-dimensional tensor using a third sampling matrix to generate a further modified three-dimensional tensor; and
reconstructing hyperspectral data from said further modified three-dimensional tensor using said sampling matrices and said third sampling matrix,
said sampling two modes of a three-dimensional tensor being performed by measuring light intensity, resulting samples being represented by digital data, and said sampling a third mode of said modified three-dimensional tensor being performed by processing said digital data.
5. A method comprising:
obtaining data representative of a scene across spectral bands using a compressive-sensing-based hyperspectral imaging system comprising optical elements;
sampling two modes of a three-dimensional tensor corresponding to a hyperspectral representation of said scene using said compressive-sensing-based hyperspectral imaging system, to generate a modified three-dimensional tensor;
after said sampling two modes, sampling a third mode of said modified three-dimensional tensor using an external processor separate from said compressive-sensing-based hyperspectral imaging system to generate a further modified three-dimensional tensor; and
reconstructing hyperspectral data from said further modified three-dimensional tensor using said external processor,
said reconstructing comprising recovering each mode of said three-dimensional tensor via a sequence of optimization processing in a reverse order of said sampling two modes and said sampling a third mode, said sequence comprising:
applying an optimization process to said further modified three-dimensional tensor to reconstruct said modified three-dimensional tensor as a recovered modified three-dimensional tensor; and
applying an optimization process to said recovered modified three-dimensional tensor to reconstruct said three-dimensional tensor corresponding to said data representative of said scene.
6. The method according to claim 5 , said applying an optimization process to said further modified three-dimensional tensor comprising performing vectorial optimization on said third mode of said further modified three-dimensional tensor, and
said applying an optimization process to said recovered modified three-dimensional tensor comprising performing joint tensorial optimization on said two modes of said recovered modified three-dimensional tensor.
7. A method comprising:
obtaining data representative of a scene across spectral bands using a compressive-sensing-based hyperspectral imaging system comprising optical elements;
sampling two modes of a three-dimensional tensor corresponding to a hyperspectral representation of said scene using said compressive-sensing-based hyperspectral imaging system, to generate a modified three-dimensional tensor;
after said sampling two modes, sampling a third mode of said modified three-dimensional tensor using an external processor separate from said compressive-sensing-based hyperspectral imaging system to generate a further modified three-dimensional tensor; and
reconstructing hyperspectral data from said further modified three-dimensional tensor using said external processor,
said sampling two modes of a three-dimensional tensor being performed by measuring light intensity, and resulting samples being represented by digital data.
8. The method according to claim 7 , said sampling a third mode of said modified three-dimensional tensor being performed by processing said digital data.
9. A system comprising:
a compressive-sensing-based hyperspectral imaging system comprising:
optical elements obtaining data representative of a scene; and
a image processor operatively connected to said optical elements,
said compressive-sensing-based hyperspectral imaging system sampling two modes of a three-dimensional tensor corresponding to a hyperspectral representation of said scene using sampling matrices, one for each of said two modes, to generate a modified three-dimensional tensor,
said image processor sampling a third mode of said modified three-dimensional tensor using a third sampling matrix to generate a further modified three-dimensional tensor after said compressive-sensing-based hyperspectral imaging system performs said sampling two modes, and
said image processor reconstructing hyperspectral data from said further modified three-dimensional tensor using said sampling matrices and said third sampling matrix,
said image processor reconstructing hyperspectral data by recovering each mode of said three-dimensional tensor via a sequence of optimization processing in a reverse order of said sampling two modes and said sampling a third mode, said sequence comprising:
applying an optimization process to said further modified three-dimensional tensor to reconstruct said modified three-dimensional tensor as a recovered modified three-dimensional tensor; and
applying an optimization process to said recovered modified three-dimensional tensor to reconstruct said three-dimensional tensor corresponding to said data representative of said scene.
10. The system according to claim 9 , said applying an optimization process to said further modified three-dimensional tensor comprising performing vectorial optimization on said third mode of said further modified three-dimensional tensor, and
said applying an optimization process to said recovered modified three-dimensional tensor comprising performing joint tensorial optimization on said two modes of said recovered modified three-dimensional tensor.
11. A system comprising:
a compressive-sensing-based hyperspectral imaging system comprising:
optical elements obtaining data representative of a scene; and
a image processor operatively connected to said optical elements,
said compressive-sensing-based hyperspectral imaging system sampling two modes of a three-dimensional tensor corresponding to a hyperspectral representation of said scene using sampling matrices, one for each of said two modes, to generate a modified three-dimensional tensor,
said image processor sampling a third mode of said modified three-dimensional tensor using a third sampling matrix to generate a further modified three-dimensional tensor after said compressive-sensing-based hyperspectral imaging system performs said sampling two modes, and
said image processor reconstructing hyperspectral data from said further modified three-dimensional tensor using said sampling matrices and said third sampling matrix,
said compressive-sensing-based hyperspectral imaging system sampling said two modes by measuring light intensity, and resulting samples being represented by digital data.
12. The system according to claim 11 , said image processor sampling said third mode by processing said digital data.
13. A system comprising:
a compressive-sensing-based hyperspectral imaging system comprising optical elements obtaining data representative of a scene, and a image processor; and
an external processor separate from and operatively connected to said compressive-sensing-based hyperspectral imaging system,
said compressive-sensing-based hyperspectral imaging system sampling two modes of a three-dimensional tensor corresponding to a hyperspectral representation of said scene to generate a modified three-dimensional tensor;
said image processor outputting said modified three-dimensional tensor to said external processor;
said external processor sampling a third mode of said modified three-dimensional tensor to generate a further modified three-dimensional tensor after said compressive-sensing-based hyperspectral imaging system performs said sampling two modes; and
said external processor reconstructing hyperspectral data from said further modified three-dimensional tensor,
said external processor reconstructing hyperspectral data by recovering each mode of said three-dimensional tensor via a sequence of optimization processing in a reverse order of said sampling two modes and said sampling a third mode, said sequence comprising:
applying an optimization process to said further modified three-dimensional tensor to reconstruct said modified three-dimensional tensor as a recovered modified three-dimensional tensor; and
applying an optimization process to said recovered modified three-dimensional tensor to reconstruct said three-dimensional tensor corresponding to said data representative of said scene.
14. The system according to claim 13 , said applying an optimization process to said further modified three-dimensional tensor comprising performing vectorial optimization on said third mode of said further modified three-dimensional tensor, and
said applying an optimization process to said recovered modified three-dimensional tensor comprising performing joint tensorial optimization on said two modes of said recovered modified three-dimensional tensor.
15. The system according to claim 14 , said joint tensorial optimization being one of parallelizable joint tensorial optimization and serial joint tensorial optimization.
16. A system comprising:
a compressive-sensing-based hyperspectral imaging system comprising optical elements obtaining data representative of a scene, and a image processor; and
an external processor separate from and operatively connected to said compressive-sensing-based hyperspectral imaging system,
said compressive-sensing-based hyperspectral imaging system sampling two modes of a three-dimensional tensor corresponding to a hyperspectral representation of said scene to generate a modified three-dimensional tensor;
said image processor outputting said modified three-dimensional tensor to said external processor;
said external processor sampling a third mode of said modified three-dimensional tensor to generate a further modified three-dimensional tensor after said compressive-sensing-based hyperspectral imaging system performs said sampling two modes; and
said external processor reconstructing hyperspectral data from said further modified three-dimensional tensor,
said compressive-sensing-based hyperspectral imaging system sampling said two modes by measuring light intensity, resulting samples being represented by digital data, and said external processor sampling said third mode by processing said digital data.