Space targets detection and recognition refer to the discriminations of space targets types, and threats for the purpose of accurately mastering the real-time space situation. For safety reasons, it is crucial to populate and maintain a catalog of orbiting objects. To detect and track space debris, telescopes and radars are typically used, resulting in multiple point measurements, which have to be processed in order to discriminate detections from clutter. This task in complex space background is one of the hot challenges in space field research for two main reasons. First, sensors can be either placed on ground stations or satellites making the target at few hundred kilometers. Thus, the distance with the sensors makes the target appear in the image plane as a weak point that corresponds only one or several pixels providing no edges, textures or shape information. Second, due to cloud clutter and other background clutter, the target is often submerged in the background with the intrinsic noise of the sensors, background noise, dark current noise and space radiation noise. The space-based images are then affected by the severe space environment. This challenging context makes weak target detection of great challenge requiring both specific sensors and target detection methods:
- Sensors like electronic fence, photometric system, satellite laser ranging system and so on, mainly focus on tracking and measuring satellites and space debris such as booster rockets and protective shields, as well as other kinds of comic flying objects, such as comets and asteroids. More precisely, infrared sensors are often employed attempting to detect small dim target. Indeed, Synthetic Aperture Radar (SAR) is a high-resolution all-weather and day and night sensor having very high resolution coherent imaging capabilities. And nowadays is an indispensable source of information in the Earth observation field. Within the next years in the future, several space borne SAR systems will be launched from different nations. Increased demand for interferometric products for precise digital elevation models reconstruction, for measuring the Earth geological displacement and velocities, the largest range swath width having the best azimuth resolution poses contracting requirements on efficient sensors system design. However, multi-modalities (visible, infrared, etc..) sensors also offer a prospective way to efficiently and robustly detect small dim targets.
- Space target detection mainly relies on target’s gray features and motion features. The first approaches are mainly based on the single frame difference between the intensity of the target and the neighborhood to suppress most of the background and clutter whilst the second employs the regularity of target motion to remove most of the isolated noise. In infrared dim moving target detection, the current leader methods are based on filters, statistics, time domain curves in time spectral data and random finite-set theory to cite a few. For the taking phase, existing approaches employ algorithms based on template matching tracking, optical flow, mean shift and particle filter to cite a few. But, all these algorithms can not fully handle all the challenging scenarios met in space-based images/videos.
- Ground target detection met similar challenges than space target detection.
The goals of this workshop are thus two-fold: 1) designing robust sensors for space target detection; and 2) proposing new algorithms that reach the requirements of space-based images. Papers are sollicited to address sensors and target detection to be applied in space applications, including but not limited to the followings:
- Space-borne detectors;
- Near-space vehicle-borne radar (NSVBR);
- Infrared dim moving target detection;
- Star point target detection;
- Air, ground, maritime target detection and tracking;
- Robust filter for target detection;
- Random finite-set theory for target detection;
- Robust subspace learning for target detection;
- Deep learning algorithms for target detection;
- Synthetic-aperture radar (SAR, PolSAR) ;
- Interferometric SAR (InSAR, PolInSAR, DinSAR);
- Persistent scatterer interferometry (PSInSAR);
- SAR tomography;
- Multi-chromatic analysis (MCA) for SAR (MCA-SAR and MCA-PolInSAR);
- SAR Pixel-tracking for radar targets velocity estimation;
- Moving target indicator for targets velocity estimation;
- Multi-modalities sensors;
- Sensor Networks.