Practical Low‐Cost Recovery of Spectral Power Distributions

Abstract

Measuring the spectral power distribution of a light source, that is, the emission as a function of wavelength, typically requires the use of spectrophotometers or multi-spectral cameras. Here, we propose a low-cost system that enables the recovery of the visible light spectral signature of different types of light sources without requiring highly complex or specialized equipment and using just off-the-shelf, widely available components. To do this, a standard Digital Single-Lens Reflex (DSLR) camera and a diffraction filter are used, sacrificing the spatial dimension for spectral resolution. We present here the image formation model and the calibration process necessary to recover the spectrum, including spectral calibration and amplitude recovery. We also assess the robustness of our method and perform a detailed analysis exploring the parameters influencing its accuracy. Further, we show applications of the system in image processing and rendering.

Physics vs. Learned Priors: Rethinking Camera and Algorithm Design for Task-Specific Imaging

“…Refocusing [31] Digital [33], [34], [35], [36], [37] Imaging Multispectral Decomposition [41], [42] Light Transport…”

“…Similarly, a multispectral image can be obtained by placing a multispectral filter array in front of the sensor, instead of a color filter array [33], [34]. A multispectral image can also be obtained by placing diffraction gratings [35], prisms [36], or compound optics [37] in front of the sensor. In a similar spirit to the Bayer filter, full Stokes imaging can be performed by using a micropolarimeter, where micropolarizers at different linear polarization orientations are placed in front of each pixel [40].…”

Optimized Multi-Spectral Filter Array Based Imaging of Natural Scenes

“…Consider a multi-spectral camera with N channels ( ). Let channel i ( ) have the spectral sensitivity function that can be obtained from specification sheets or measured using a broadband light source using some low-cost methods [ 29 ]. Let the response of channel i of the camera to be .…”

Practical Capture and Reproduction of Phosphorescent Appearance

“…In concurrent work, Alvarez-Cortes et al [ACKM16] propose to use diffraction filters to image spectral distributions of light sources at low cost while we are using linear variable bandpass filters (LVFs) for this purpose. While pictures taken using these two types of filters both show "images" of the spectrum the spatial resolution is typically much higher for an LVF which is beneficial as we are operating at very low intensity levels and averaging pixels corresponding to the same wavelength helps increasing robustness.…”