Local mass distribution partitioning for object recognition
US7388990B2 · kind B2 · utility
Assignee
Inventors
Key dates
| Filing date | Sep 22, 2003 |
| Grant date | Jun 17, 2008 |
| Priority date | — |
| Expiry date | Jan 1, 2026 |
Classification
- Technology area (CPC G)Physics
- CPC primaryG06V30/242
- WIPO fieldComputer technology
- WIPO sectorElectrical engineering
Abstract
A method for determining a similarity score of a target object with respect to a model object. The target object is in a plane and the model object is represented by a model feature vector. The method comprises generating regions of the plane according to a first mass distribution of the target object and a second mass distribution of a part of the target object. Each of the generated regions has a corresponding mass distribution indicator. The method further comprises calculating a target feature vector for the target object according to at least one of the corresponding mass distribution indicators. Finally, the method computes the similarity score using the target feature vector and the model feature vector.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.