|Title||SPIN: Cleaning, Monitoring, and Querying Image Streams Generated by Ground-Based Telescopes for Space Situational Awareness|
|Publication Type||Journal Article|
|Year of Publication||2017|
|Authors||Keke Chen, Bharath Avusherla, Sarah Allison, Vincent Schmidt|
|Keywords||Data Stream, Image Processing, Image Query Processing, machine learning, Space Situational Awareness|
With the increasing number of objects in Earth orbits, space situational awareness (SSA) becomes critical to space safety. As an economical option, ground-based telescopes can be deployed around the world and continuously provide imaginary information of space objects. However, they also raise unique challenges regarding big, noisy, and streaming data processing. In this paper, we present the SPIN system to address these challenges. The core algorithms process image sequences generated by ground-based telescopes and conduct: (1) image quality classification for data cleaning, (2) stream-based key-object identification and anomaly detection, and (3) efficient query processing on large image sequence repositories. Our goal is to design or adopt algorithms that handle the domain-specific image streams most efficiently and effectively. We use a 17-inch telescope to collect a large real dataset for evaluating the core algorithms, which covers more than ten satellites in one month and contains about 16,400 images. The experimental results show that the developed algorithms are fast enough for stream based real-time processing and also yield high-quality results for all the primary tasks.
|Full Text Citation|| |