Document authentication using template matching with fast masked normalized cross-correlation
View Patent ↗Techniques are described for authenticating security documents having security images that incorporate multiple security features. The techniques may produce robust template matching results in situations of lighting unevenness, even with stronger occluding objects. The techniques may be particularly useful in validating a security document having a security image composed of one or more “virtual” retroreflective images formed over a background of a repeating retroreflective image. The virtual retroreflective images within the security image provide strong signals that may dominate analysis and validation of the background retroreflective image, thereby resulting incorrect authentication. The techniques provide a modified normalized cross-correlation analysis that masks out the strong signals contributed by the additional security features, such as the retroreflective virtual images, while still allowing for fast and efficient template matching to be performed with respect to the background image.
1. A method comprising:
capturing an image of at least a portion of a document, wherein the portion of the document includes an image having a first retroreflective feature and a second retroreflective feature;
generating a masking image from the captured image;
calculating a modified normalized cross-correlation between a reference image associated with the first retroreflective feature and the captured image using the masking image; and
outputting indicia of one or more matches between the reference image and the captured image.
2. The method of claim 1 , wherein calculating the modified normalized cross-correlation between the reference image and the captured image comprises:
applying the masking image to the captured image to obtain a masked image;
calculating a normalized cross-correlation between the masked image and the reference image; and
modifying the calculated normalized cross-correlation with a masking effect modifier representing the ratio of the size of the reference image to a masked area at each matching region.
3. The method of claim 2 , wherein the masking effect modifier is of the form
{
{
M
t
×
N
t
}
∑
x
,
y
m
(
x
+
u
,
y
+
v
)
}
α
,
where M t and N t define the dimension of the reference image, m(x, y) represents the masking image, and α represents the modification strength.
4. The method of claim 3 , wherein modifying the calculated normalized cross-correlation with a masking effect modifier comprises modifying using a masking effect modifier in which α=1.
5. The method of claim 2 , wherein the second retroreflective feature is a retroreflective virtual image, and wherein applying the masking image to the captured image removes at least a portion of the retroreflective virtual image from the captured image.
6. The method of claim 2 ,
wherein the masking image and the captured image are comprised of a plurality of pixels, and
wherein generating a masking image comprises setting a pixel of the masking image to a first value when a corresponding pixel of the captured image has a brightness value outside a predefined range of brightness values, and setting the pixel of the masking image to a second value when the corresponding pixel of the captured image has a brightness value within the pre-defined range of brightness values.
7. The method of claim 6 , further comprising selecting the predefined range of brightness values from among a plurality of predefined ranges of brightness values based on a type of the document.
8. The method of claim 1 , further comprising determining matching regions within the captured image based on the calculated modified normalized cross-correlation.
9. The method of claim 8 , wherein outputting indicia of one or more matches comprises displaying the determined matching regions and a correlation score corresponding to each determined matching region.
10. The method of claim 8 , wherein outputting indicia comprises indicating a failure to find any matching regions within the captured image, wherein a failure to find any matching regions is determined based on a predefined score threshold and a minimum required number of matches.
11. The method of claim 1 , wherein calculating the modified normalized cross-correlation between the reference image and the captured image comprises calculating the modified normalized cross-correlation using a fast Fourier transform.