Image Processing Reference
In-Depth Information
Fig. 10.19. Minutia detection process. Adapted after [76]
Chap. 11. Image V represents a fusion of the information in image III and IV by
pointwise multiplication according to (
IV
)
). Finally, image VI presents
the automatically found minutiae. The positions of the minutiae as well as the di-
rections between and around them are useful personal attributes in biometric person
authentication. The details of this minutiae detection approach are given in [76].
·
(1
−|
III
|
Image Enhancement
Image enhancement is a valuable tool to reinforce the quality of recoverable details.
In Fig. 10.20, we show the result of line reinforcement in color images [224]. The
processing performs a so-called shock filltering to bias lines by means of direction
estimation using the structure tensor.
10.13 Discrete Structure Tensor by Spectrum Sampling (Gabor)
We will attempt to estimate the
discrete structure tensor
by
spectrum sampling
as fol-
lows. In Sect. 9.6 we concluded that the responses of a set of Gabor filters constitute
a sampled version of the local Fourier spectrum. Each filter is tuned to a particular
position in the spectral domain corresponding to a particular frequency and direction
with a certain support in the frequency domain. Assuming uniform sampling, the
