Testing spatial modulations in a compressive single-pixel hyperspectral microscope with optical fiber light collection

Abstract A single-pixel camera offers an alternative method for image acquisition by using a series of modulation patterns on the scene. When combined with compressive sensing, it can significantly reduce the number of required measurements to reconstruct the image. Various modes of spatial modulations in compressive imaging can vastly affect the measurement quality and reduce the number of measurements needed. While previous studies have explored various modulation approaches, they often examined only a limited subset under inconsistent conditions. In this work, we provide a thorough comparison of the performance of 20 different modes of modulations and their modifications using a unified testing scheme. These were implemented via a digital micromirror device (DMD), and the DMD-modulated light was collected into an optical fiber. First, we evaluated the modulation patterns through simulations to study the impact of noise on image reconstruction. Next, we experimentally tested the entire set of patterns using the previously developed dual hyperspectral compressive microscope. We observed that the experimental results were highly affected by the diffraction on DMD, which highly depended on the character of a pattern. To ensure a thorough analysis, all modulation modes were tested on three different samples and at three distinct resolutions. Additionally, we identified individual challenges and limitations associated with specific modulation modes, offering practical insights for their testing and application. This study provides valuable guidance for optimizing modulation patterns in compressive imaging systems.