Modeling and Information Processing
DOI - Detecting Objects of Interest
Objective
- To design data processing algorithms for the automatic extraction of uncooperative objects from data from physical sensors.
Contribution
The detection and geometric location of objects is a prerequisite for a large number of interpretation and decision making functions because it provides information about the existence of entities of operational interest.
The DTIM is interested in two types of problems:
- Developing algorithms integrating the operational constraints and knowledge specific to a target aerospatial context. The objective of the work is either to evaluate the ability of documented algorithms to satisfy the specific requirements or to design innovative approaches that provide a response to particular contexts of use. Statistical estimation, geometric reasoning and learning are some of the basic techniques used.
- Modeling performance. This challenging stage is difficult but necessary for the overall design of systems including, amongst other things, a detection function. Generally, it uses the statistical modeling of signals.
Illustrations
- Detecting weak objects in a clutter sequence
- Detecting areas of interest in an SAR image
- Detecting weak objects from complementary multiphysics sensors
- Detecting complex objects in very large images
