Implicit Neural Representation for Hyperspectral Video Compression
The paper reports large compression gains for hyperspectral video using an implicit neural representation model.
The authors extend an RGB video compression approach to hyperspectral video, comparing it with frame-by-frame hyperspectral image compression. They report +4.99 dB BD-PSNR and an 88.88% BD-rate reduction. They also test downstream tracking, where the method improves HOT2026 object-tracking metrics in low-data settings against PCA and JPEG2000-based compression. Accepted at IEEE WHISPERS 2026. ArXiv · AI/CL/LG's note
The authors extend an RGB video compression approach to hyperspectral video, comparing it with frame-by-frame hyperspectral image compression. They report +4.99 dB BD-PSNR and an 88.88% BD-rate reduction. They also test downstream tracking, where the method improves HOT2026 object-tracking metrics in low-data settings against PCA and JPEG2000-based compression. Accepted at IEEE WHISPERS 2026. ArXiv · AI/CL/LG's note
score 3