Spectral Security Imaging System (SSIS): Optical Authenticity for Hyperspectral Pushbroom Imaging
Pablo Andres Gomez Toloza, Javier Andres Quiroga Torres, Hans Garcia, Gonzalo Arce, Henry Arguello Fuentes
AI summary
Problem
Post-acquisition digital authentication cannot guarantee the authenticity of hyperspectral imagery because subtle spectral manipulations can occur before digitization. Existing optical methods often lack complete hardware implementation or use multiplicative keys that degrade downstream classification performance.
Approach
The SSIS prototype physically embeds a learned, additive spectral key into the pushbroom imager's optical path, binding data integrity to the analog acquisition process. A joint optimization framework trains the key and a detection network to maximize manipulation detection while minimizing distortion to class-discriminative spectral features.
Key results
- 92% manipulation detection accuracy with PSNR of 41.5 dB and SSIM of 0.981
- Classification macro-F1 retains 93% of monochromator and 99% of pushbroom baselines
- Outperforms multiplicative and watermarking baselines by up to 16.1 macro-F1 points
- Complete laboratory calibration yielding 51 separable spectral bands across 400–850 nm
Why it matters
Critical for remote sensing, environmental monitoring, and robotic perception systems that rely on unaltered hyperspectral data for accurate decision-making.
Abstract
Ensuring authenticity of hyperspectral imagery (HSI) at the moment of acquisition is critical: subtle spectral attacks can mislead downstream analysis before digital defenses take effect. By injecting the optical key before digitization, the effect is created in hardware and cannot be replicated in software, resulting in stronger protection. We present the Spectral Security Imaging System (SSIS), an acquisition-stage approach that injects a data-aware additive spectral key in the optical path of a pushbroom imager, binding integrity to the measurement while preserving the class-informative structure. We describe the complete system forward model, detector–key joint optimization, and a laboratory prototype together with a thorough calibration process. A laboratory dataset (unsigned and optically signed cubes) supports two evaluations. For manipulation detection, SSIS achieves detection accuracy of 92% with visual distortion of PSNR = 41.5 dB and SSIM = 0.981. For downstream classification on clean data, macro- F1 remains close to the unsigned ceiling, about 93% of the monochromator baseline (0.915 vs 0.981) and about 99% of the pushbroom baseline (0.892 vs 0.903), while outperforming multiplicative and watermarking baselines by up to 16.1 points in macro-F1 and 19.4 points in accuracy.