Palmprint alignment using intrinsic local affine-invariant fiducial points residing on area patches spanning between two successive fingers

This paper deals with the problem of palmprint alignment based on intrinsic local affine-invariant fiducial points residing on the area patches spanning between two successive fingers. The palmprint data is acquired by using general USB camera. Because these fiducial points are local and intrinsic, they allow for partial alignment, where only some part of the palmprint is viewed. Moreover, since the fiducial points are relative affine invariant to affine transformations, they allow for alignment where position of the palm relative to camera orientation can be arbitrary set. A fast non-iterative alignment procedure is proposed that establishes reliable correspondences between fiducial points without any prior knowledge of the overall global transformation that took place after the changes in camera orientation. This is achieved through the construction of a set of ordered novel absolute local affine invariants. With enough fiducial points set as correspondents, the overall affine transformation is computed and the palmprint before and after the transformation are aligned. The average alignment error was found to be in the order of 0.5% of the size of the palm corresponding to the physical resolution of 2.6 mm.