Abstract
Across optics and photonics, excess intensity noise is often considered a liability. Here, we show that excess noise in broadband supercontinuum and superluminescent diode light sources encodes each spectral channel with unique intensity fluctuations, which actually serve a useful purpose. Specifically, we report that excess noise correlations can both characterize the spectral resolution of spectrometers and enable cross-calibration of their wavelengths across a broad bandwidth. Relative to previous methods that use broadband interferometry and narrow linewidth lasers to characterize and calibrate spectrometers, our approach is simple, comprehensive, and rapid enough to be deployed during spectrometer alignment. First, we employ this approach to aid alignment and reduce the depth-dependent degradation of the sensitivity and axial resolution in a spectrometer-based optical coherence tomography (OCT) system, revealing a new outer retinal band. Second, we achieve a pixel-to-pixel correspondence between two otherwise disparate spectrometers, enabling a robust comparison of their respective measurements. Thus, excess intensity noise has useful applications in optics and photonics.
Highlights
Across optics and photonics, excess intensity noise is often considered a liability
B, NA,out(xA) and NB,out(xB), respectively, are cross-correlated to yield RAB,out(xA, xB), where the highest correlation values occur for pixels that measure similar wavelengths broadband light with high-resolution spectral encoding, which is a natural conduit for spectrometer characterization and cross-calibration
We demonstrate its utility by improving the spectral resolution of multiple visible light optical coherence tomography (OCT) spectrometers and visualizing a new band in the mouse outer retina
Summary
Spectrometer characterization, essentially reduces to system identification[7,8]. The first, the impulse response method, determines the spectral resolution from the measured intensity pattern of a narrow linewidth light source or sharp spectral feature, ideally with a lineshape much narrower than the spectral resolution (Fig. 1b). This approach requires additional narrow linewidth sources, narrowband optical filters, or sources with fine spectral features (e.g., Fraunhofer lines or frequency combs). The second, the transfer function method, determines the spectral resolution across wavelengths from the attenuation of a sinusoidal interference fringe pattern envelope as the path length mismatch increases (Fig. 1c). Shot noise must be uncorrelated between pixels, with a variance proportional to the pixel gray level, while the excess noise variance goes as the square of the gray level
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have