This paper presents a novel imaging spectropolarimeter that acquires the spatial, spectral, and polarization information of a scene without scanning in the spatial or spectral domains. Current approaches include combining channeled spectropolarimetry with an image mapping spectrometer, using Fourier-transform-based processing. The disadvantages of Fourier processing are noise sensitivity, channel cross-talk, and resolution limitations due to filtering. Instead, this work builds a coded aperture snapshot spectral imager to capture spatial, spectral, and polarization information based on compressive sensing. The optical system modulates the Stokes parameters of the incident light. A coded aperture encodes the four-dimensional information of the object, followed by a double Amici prism that disperses the encoded datacube. A camera measures images as a polarized filter wheel rotates through five positions. The inverse problem is to reconstruct the four data cubes for each of the Stokes parameters from the five camera measurements. The theory of compressive sensing provides visually pleasing reconstructions by using so-called L1 techniques or total variation minimization. This paper demonstrates an experiment that acquires Stokes parameter images in the visible wavelengths. It alleviates channel crosstalk and resolution limitations of traditional channeled spectropolarimeters and reduces the number of measurements for imaging spectropolarimetry.

---

---

 Add Comment

You must log in to add comments.