Needles in a Haystack: Tracking City-Scale Moving Vehicles from Continuously Moving Satellite

Area of Science:

• Computer Vision
• Remote Sensing
• Image Analysis

Background:

• Satellite videos capture city-scale scenes with tiny moving targets.
• Challenges include small target size (few pixels), noise, and uneven background motion due to satellite movement.

Purpose of the Study:

• To propose a novel framework for identifying small moving vehicles in satellite videos.
• To address the limitations of existing technologies in detecting such tiny targets.

Main Methods:

• A novel detection algorithm based on local noise modeling.
• Differentiating targets from noise using an exponential probability distribution.
• A multi-morphological cue-based strategy for further noise discrimination.

Main Results:

• The proposed framework successfully identifies small moving vehicles in satellite videos.
• Evaluation protocols were developed and used to systematically measure performance.
• The algorithm demonstrated advantages over state-of-the-art baselines.

Conclusions:

• The novel framework offers an effective solution for tiny moving vehicle detection in satellite imagery.
• The developed evaluation protocols provide a standardized method for performance measurement.
• The approach advances the field of computer vision for remote sensing applications.