Computational hyperspectral imaging using diffractive phase masks
Abstract
Objective:The objective of this project is to develop a lightweight, compact hyperspectral imager that can:(1) adaptively image within specified spectral bands with either minimal changes in hardware orvia software modifications alone and(2) image such that the resolution in the spatial domain can be traded off against that in the spectral domain essentially via parametric adjustments in the computational algorithms.Approach:We design a multi-~~level diffractive micro-~~optic (that we call the polychromat) such that the dispersed image created on the sensor array is unique for each wavelength-~~spatial-~~frequency (~-~~k) combination. During the calibration step, we will calibrate the image formed at each ~-~~k combination and create a lookup table. Note that these wavelength samples can be in one continuous band or in multiple non-~~contiguous bands. We can readily design the polychromat to operate most efficiently over multiple non-~~contiguous bands of interest. Furthermore, we can trade-~~off spectral resolution against spatial resolution by adjusting the parameters in the reconstruction algorithm. The fundamental limitation, of course, comes from the channel capacity of the system. On the polychromat side, we have even more degrees of freedom to engineer the dispersive nature of the image. Assuming polychromat pixel size of 3~m (which is easily fabricated) over an area of 12mm X 12mm and 16 multiple height levels, we have a total of 256M independent degrees of freedom. Therefore, extraction of the required spectrum as well as the hyperspectral or multi-~~spectral image should be feasible. Note that we collect the entire hyperspectral (or multi-~~spectral) lightfield, Iin. This means that we can refocus the data in post-~~processing to gain 3D scene information as required. If only the far-~~field of a scene is desire, the fourier coefficients themselves correspond to the scene (with appropriate scaling factors). Statement of Work The research will proceed along the following tasks.1. Design of the camera for visible spectrum. [month 1 to 6] 2. Assembly and characterization of visible-~~spectrum camera. [month 7 to 18] 3. Design of camera for SWIR spectral range. [Month 6 to 18] 4. Fabrication of polychromat for SWIR camera [Month 18 to 24] 5. Assembly and characterization of SWIR camera. [Month 24 to 36] 6. Massively parallel reconstruction algorithms. [Month 1 to 36 (ongoing)] 7. Implementation in embedded hardware [Month 1 to 36] (ongoing)
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 25, 2017
- Source ID
- N000141512316
Entities
People
- Rajesh Menon
Organizations
- Office of Naval Research
- United States Navy
- University of Utah