| 摘要 |
For a conventional object image recognition technique, recognition with high reliability is possible since a large number of features are extracted from an object, but redundant features and duplicated features can also be increased. In addition, a time for object search prolongs as the features increases. In an embodiment of the present invention, provided are an apparatus and a method for object recognition which are strong for the object′s pose, size, and occlusion and are capable of reducing a matching time for the object recognition. [Reference numerals] (100) Boundary extracting unit; (200) Property extracting unit; (300) Post-processing unit; (AA) Input an object image |
