摘要 |
Disclosed are a method and an apparatus to recognize an object. The object recognizing method includes the steps of: extracting feature points of an object to be recognized from an input image; generating a feature image through normalization by producing direction information of an adjacent area based on the feature points; producing a covariance descriptor based on the feature image; and recognizing an object from the input image by comparing the covariance descriptor with a prerecorded reference covariance descriptor. Thus, the present invention is able to improve the recognition performance of an object by recognizing the object at a fast speed while being strong against the change of an image such as size, rotation, and illumination. |