Hyperspectral video compression improves quality and tracking accuracy
Implicit Neural Representation for Hyperspectral Video Compression
Computer Vision and Pattern RecognitionArtificial IntelligenceMachine Learning
Summary
Hyperspectral videos capture more colors than regular videos but create huge data files that are hard to store and share. The authors improved how these videos are compressed using a special kind of neural network, making the files much smaller while keeping the video quality better than older methods. This also helped computer programs track objects in the videos more accurately. Their method works better, especially when there isn’t much data to work with.
What this means in practice
- •For remote sensing teams: Compress large hyperspectral video datasets efficiently while maintaining quality for better object tracking in surveillance and environmental monitoring.
- •For satellite imaging operators: Improve storage and transmission of hyperspectral videos from satellites by reducing data size without losing important tracking details.
Authors
Alfredo Scalera, Paul Murray, Jaime Zabalza
Abstract
With the advent of snapshot cameras, hyperspectral video is becoming more readily available. In recent years, new applications have emerged which have led to increasingly larger datasets. However, hyperspectral video compression remains in the early stages. In this study, we explore the use of implicit neural representation as a candidate solution. We propose a novel extension of an existing RGB video compression model, achieving Bjøntegaard Delta PSNR gains of +4.99 dB and Bjøntegaard Delta rate of -88.88% compared to traditional hyperspectral image compression methods applied frame-by-frame. In addition to reconstruction quality, the effects on downstream task performance are measured in the form of object tracking success. Compared to video compressed with methods based on principal component analysis and JPEG2000 in low data regimes, our proposed method improves tracking area under the curve by up to 23.42% and distance precision by up to 35.56% on examples from the HOT2026 dataset.