Embedded Files
The Limitation of Aerial Imagery
Aerial imagery is widely used for surveillance and reconnaissance, yet unstable atmospheric conditions like fog and clouds significantly impact image quality. Due to the nature of aerial photography, especially high-altitude wide-angle shots, atmospheric transmission is substantially reduced, leading to inevitable image degradation from atmospheric scattering.
How to improve the system?
To improve Aerial ISR, I developed a machine-learning based system:
1) Enhance the usability of aerial imagery by proposing a real-time fog reduction technique.
2) Detect imagery anomalies by AI-enhanced Image Quality Monitoring.
1. Aerial Video Dehazing using Dark Channel Prior Algorithm
The light reflected from an object and the light from an external light source scattered in the atmosphere, is referred to as fog. The degradation of the image due to atmospheric scattering can be modeled from the distance from the camera and the density of atmospheric particles.
2. Anomaly Detection on Video Quality Assessment
Real-time image quality anomalies in complex atmospheric conditions (fog, clouds) are detected by monitoring time-series data using no-reference metrics (BRISQUE, NIQE, Blur). Multivariate anomalies are identified through Isolation Forest analysis.
Page updated
Google Sites
Report abuse