Objective
Research objectives and content The proposed research project centres on combining images from different views in order to synthesise new images which represent changes in the scene under rigid 3D transformations. The research team at Gravir-Inria has already developed the geometrical tools needed in the synthesis process when the images are obtained via an uncalibrated camera (for instance from a handycam manipulated without a special car). An important issue in the synthesis process is to obtain sufficient and accurate matching points from two images. The proposed project will focus on the problem of computing these points. The aim is to formulate a robust strategy based on the multiple influence between potential matching points. This formulation will search for a global consistent solution and will consider the problem of matching reliability (i.e. wrong matching hypothesis) which emerges due to occlusion and viewpoint changes. These changes can radically alter the appearance of local features reducing significatively the efficacy of the matching. The formulation developed must provide accurate results for image synthesis constituting a valuable tool in areas such as visual inspection for the design of objects and industrial parts, in medical diagnosis and in the visualisation of virtual environments. Additionally, an accurate matching process can have a fundamental role in many other applications such as the automatic manipulation of industrial parts, autonomous vehicle guidance as well as target identification. Training content (objective, benefit and expected impact) The research group at Grenoble has a great experience in the modelling of the geometry of multiple- views and 3D reconstruction. A stay there would allow me to acquire a deep knowledge and experience with one of the leading groups in geometry invariance and applications in 3D vision. My research background in the fields of image and signal processing would contribute to the development of new solutions to complex problems in the area of 3D vision. | Links with industry / industrial relevance (22) The research group at Grenoble has links with Getris Image which is interested in virtual studio features as well with Matra which founded part of the preliminary study of this project (further funding is under consideration). The successful results of the proposed research project would allow low cost realistic image synthesis for very different markets such as home entertainment, industrial or medical simulations.
Fields of science (EuroSciVoc)
CORDIS classifies projects with EuroSciVoc, a multilingual taxonomy of fields of science, through a semi-automatic process based on NLP techniques.
CORDIS classifies projects with EuroSciVoc, a multilingual taxonomy of fields of science, through a semi-automatic process based on NLP techniques.
- engineering and technologymechanical engineeringvehicle engineeringautomotive engineeringautonomous vehicles
- engineering and technologyelectrical engineering, electronic engineering, information engineeringelectronic engineeringsensorsoptical sensors
- engineering and technologyelectrical engineering, electronic engineering, information engineeringelectronic engineeringsignal processing
- natural sciencesmathematicspure mathematicsgeometry
You need to log in or register to use this function
Call for proposal
Data not availableFunding Scheme
RGI - Research grants (individual fellowships)Coordinator
38330 BIVIERS
France