Self-calibrated neuromorphic hyperspectral derivative imaging

Hyperspectral imaging provides rich spectral-spatial contrast, but conventional systems are often limited by cost, acquisition speed, and dynamic range. We present OpenHI, a low-cost event-based hyperspectral imager that uses an event camera during spectral scanning to capture spectral-derivative frames with respect to the wavelength of the scene’s log-transmittance, achieving microsecond latency and high dynamic range. To address distortions from non-constant scan velocity, including acceleration-induced temporal shear, we apply a multi-window warping model for compensation. The system also automatically recovers scan timing and performs time-wavelength alignment. We compare reconstructed spectral slices with a reference hyperspectral camera under low-light conditions and demonstrate an open, low-cost hardware–firmware–software pipeline that makes neuromorphic hyperspectral imaging more accessible.