Method, apparatus and computer program for tracking of moving objects
The invention concerns the correct tracking of several objects, which can be seen in a video, and the simultaneous, correct identification of the objects. This means which individual is involved.
Existing camera-based people tracking systems produce false trajectories when people are obscured by other people or obstacles, or if the objects look very similar. Especially in the sports sector, people tracking systems are not able to provide correct trajectories.
By using the inertial measurement unit, the invention should allow correct functioning even in complex scenarios and supply the correct trajectories. In addition, the invention should enable each trajectory to be automatically assigned to the corresponding object. The invention works with a minimal number of one inertial measuring device per person. If the identification of objects by the invention works independently of the visual appearance, the invention can also be used to identify, locate and track objects in night images (near infrared). This is an absolute novelty.
- The invention can lead to a cost reduction in player analysis in the sports sector.
- The invention can be used to analyze the customer base of a shopping center very accurately.
- The invention allows residents of a nursing home to be equipped with inertial measurement units so that they can be automatically localized and monitored for their own protection.
- The invention can also be applied to animals (if an inertial measuring unit can be fitted). In particular, it can be used to analyse herd and swarm behaviour. Here it comes very often to occlusions. In addition, it is difficult even for humans to keep very similar looking animals apart. Accordingly, all existing video-based tracking systems produce false results. The invention makes it possible for the first time to analyze individual behavior within a swarm of bees or a pack of wolves
Scope of application
The field of application is image processing.
License for commercial exploitation / Research & Development cooperation
Proof of concept
- WO 2020/239210 A1 (angemeldet in DE) pending
KeywordsGottfried Wilhelm Leibniz University Hannover, identification, image processing, Inertial measuring instrument, Measuring instrument, Tracking, Trajectory
Offer at Providers website