| 摘要 |
A method ( 500 ) of generating a matching key ( 599 ) for an image ( 501 ) is disclosed. The matching key ( 599 ) is substantially invariant to rotation, scale and translation. The method ( 500 ) starts by forming a spatial domain representation of the image ( 501 ) that is substantially invariant to translation of the image ( 501 ). Rotation and/or scaling in the spatial domain representation is next transformed ( 580 ) into translation to form a transformed image. A representation of the transformed image is then formed that is substantially invariant to translation of the transformed image. The representation of the transformed image is the matching key ( 599 ) for the image ( 501 ).
|
