Method and system for daytime infrared space surveillance
A space surveillance method for detecting space objects in orbit around the Earth in images captured during the daytime, the method including the following steps: capturing a plurality of infrared images of the daytime sky using a camera including at least one infrared sensor, detecting space objects in orbit around the Earth on the basis of the images, the detection of bright spots being implemented by a deep-learning artificial intelligence system, and identifying each object detected from a catalogue of known space objects in orbit around the Earth.
1 . A space surveillance method for detecting space objects in orbit around Earth in images captured during daytime, the method comprising:
capturing a plurality of infrared shots of a daytime sky using a camera comprising at least one infrared sensor, each infrared shot comprising an array of pixels which are each associated with an intensity of light received by an infrared sensor,
detecting space objects in orbit around the Earth on a basis of said shots,
identifying each object detected from a catalogue of known space objects in orbit around the Earth,
the method being characterised in that the step of detecting space objects in orbit around the Earth is implemented by a deep-learning artificial intelligence system comprising a plurality of layers of artificial neural network connected together in order to analyse information from a preceding layer of neurons, the deep-learning artificial intelligence system being based on simulation images generated in order to reproduce typical images coming from the infrared sensor and comprising a background and a background noise, spots of light corresponding either to stars or to space objects or to defects of the infrared sensor, each simulation image being associated with a truth based on positions of real objects in the image,
and the step of detecting space objects in orbit around the Earth comprising:
detecting bright spots in each shot,
discriminating the detected bright spots, the discrimination comprising tracking each detected bright spot that is stationary in successive shots, and recording coordinates of the detected bright spots at possibly different positions and grouped together by this tracking, the recording being performed, for each bright spot detected, following its disappearance in the following shots.
2 . The method according to claim 1 , wherein the layers of artificial neural networks are calibrated, prior to their use for detecting space objects, by a supervised learning method, using a base of various images enabling the artificial intelligence system to determine typical features of a space object.
3 . The method according to claim 1 , further comprising, immediately after capturing shots, applying a non-uniformity correction to the captured shots.
4 . The method according to claim 1 , further comprising filtering each shot.
5 . The method according to claim 1 , further comprising forming stacked images from a superposition of a plurality of said shots, each pixel of a stacked image being associated with a received intensity of light corresponding to an average of the intensities of the superimposed shots for the same pixel, the detection of space objects using the stacked images as shots to be processed.
6 . The method according to claim 5 , further comprising, before the step of detecting the bright spots, a destriping step of each stacked image in order to remove streak defects in the stacked image.
7 . A space surveillance system for detecting space objects in orbit around Earth, the system comprising a reflecting telescope mounted on a mechanical support with motorised displacement, a camera comprising at least one infrared sensor mounted at an output of the reflecting telescope and configured to take series of shots of a daytime sky at a frequency between 1 Hz and several hundred Hertz, and a processing unit receiving each shot captured by the camera,
characterised in that the processing unit comprises a deep-learning artificial intelligence system comprising a plurality of layers of artificial neural network connected together to analyse information of a preceding layer of neurons, the processing unit being configured to carry out the following steps based on received images:
detecting space objects in orbit around the Earth on a basis of the captured shots, the detection of bright spots being performed by the deep-learning artificial intelligence system, and
identifying each object detected from a catalogue of known space objects in orbit around the Earth,
the detecting of space objects in orbit around the Earth comprising detecting bright spots in each shot, and discriminating the detected bright spots, the discrimination comprising tracking each detected bright spot that is stationary in successive shots, and recording coordinates of the detected bright spots at possibly different positions and grouped together by this tracking, the recording being performed, for each bright spot detected following its disappearance in the following shots.
8 . The space surveillance system according to claim 7 , wherein the camera further comprises at least one visible light sensor mounted at the output of the reflecting telescope and configured to take series of shots of a night sky, the space surveillance system further comprising a day/night alternation module making it possible to change a type of sensor receiving the light from the sky as a function of an environmental light intensity.