Space Debris Detection in Multi-Object Tracking
- Paper number
IAC-17,A6,7,7,x38404
- Year
2017
- Abstract
This paper applies convolutional filters in multiple space debris detection and tracking. The filters are learned in both a supervised and unsupervised source from datasets through a hierarchical artificial neural network model which is a natural adaptation to such problems. Here a track-before-detect approach is employed in multiple space debris tracking, where the tracking leads the detetion. The target debris detection provides an input calculated by selecting from source features obtained by a combination of transfer and active learning, as the source context is likely to be different from target context and pricise labelling of training data is expensive. Finally, we present a simulation of these convolutional filters using a real-world dataset that verifies their performance.
- Abstract document
- Manuscript document
(absent)