Enhanced Software for Image Feature Detection and Tracking

BACKGROUND The division of Imaging Science and Biomedical Engineering at Manchester has expertise in the statistical analysis of image data. Image analysis is widely employed in sectors such as life sciences, security, defence, biometrics, forensics and medicine. THE TECHNOLOGY A method for modelling deformable objects has been developed (e.g. faces, hands, organs. The approach is general and involves building a statistical model of an image of a particular object e.g. a body part through a manually labelled set of training images. The multi-stage method automatically locates features of an image and looks at the relative position of different feature points. Algorithms, capable of matching an image against the model, use a set of feature templates and shape constrained search techniques to correlate with the target image (shape optimised search). The model is first trained to recognise a particular set of features. The search method considers the match quality of each model generated and the statistical shape constraint. This iterative search procedure returns candidate locations for each of the features in the image. The algorithm can store the model appearancefrom previous searches, making the method highly suitable for tracking objects over time. Equally, the method is suitable for static data. Some uses of the method are:
• Feature localisation in static images (e.g. photographs)
• Feature tracking in temporal data (e.g. video)
• Generalised model matching of deformable objects (e.g. medical images)
• The search algorithm is efficient and the feature points found can be used to initialise further processes. For example, classification or recognition of a class subtype (e.g. human identity, medical disease). KEY BENEFITS
• Accurate results compared to current state of the art methods (e.g. AAM – Active Appearance Model)
• Method suitable for tracking as model approximation to object appearance is updated over time
• Efficient real-time processing (~100ms per frame for video using 3.0GHz processor)
APPLICATIONS
• Avatar animation (computer games, internet forums, films, etc.)
• Face recognition and expression systems
• Fatigue detection
• Potential use in medical diagnostics INTELLECTUAL PROPERTY Expertise; considerable know-how; patent application. OPPORTUNITY Of interest to companies in the computer software, internet games, film and video animation, medical diagnostics, life sciences, biometrics, forensics, security and defence industries

Type of Offer: Licensing



Next Patent »
« More Software Patents

Share on      


CrowdSell Your Patent