Tracking of streaking targets in video frames
| Autor: | Andrew R. Finelli, Yaakov Bar-Shalom, Peter Willett, David K. Melgaard, Raymond H. Byrne |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
020301 aerospace & aeronautics
Engineering Pixel 010505 oceanography business.industry Matched filter Streak 02 engineering and technology Tracking (particle physics) 01 natural sciences Streaking Cardinal point 0203 mechanical engineering Metric (mathematics) Computer vision Artificial intelligence Noise (video) business 0105 earth and related environmental sciences |
| Zdroj: | 2017 IEEE Aerospace Conference. |
| DOI: | 10.1109/aero.2017.7943640 |
| Popis: | A method for tracking streaking targets (targets whose signatures are spread across multiple pixels in a focal plane array) is developed. The outputs of a bank of matched filters are thresholded and then used for measurement extraction. The use of the Deep Target Extractor (DTE, previously called the MLPMHT) allows for tracking in the very low observable (VLO) environment common when a streaking target is present. A definition of moving target signal to noise ratio (MT-SNR) is also presented as a metric for trackability. The extraction algorithm and the DTE are then tested across several variables, including trajectory, MT-SNR, and streak length. The DTE and measurement extraction process performs remarkably well in this difficult tracking environment on these data features. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|