Signal Processors and Methods for Estimating Geometric Transformations of Images for Digital Data Extraction

Patent metadata

Field	Value
Publication number	US-2017193628-A1
Application number	US-201615211944-A
Country	US
Kind code	A1
Filing date	Jul 15, 2016
Priority date	Jul 16, 2015
Publication date	Jul 6, 2017
Grant date	—

Abstract

Signal processing devices and methods estimate a geometric transform of an image signal. From a seed set of transform candidates, a direct least squares method applies a seed transform candidate to a reference signal and then measures correlation between the transformed reference signal and an image signal in which the reference signal is encoded. Geometric transform candidates encompass differential scale and shear, which are useful in approximating a perspective transform. For each candidate, update coordinates of reference signal features are identified in the image signal and provided as input to a least squares method to compute an update to the transform candidate. The method iterates so long as the update of the transform provides a better correlation. At the end of the process, the method identifies a geometric transform or set of top transforms based on a further analysis of correlation, as well as other results. Phase characteristics are exploited in the process of updating coordinates and measuring correlation. The geometric transform is used as an approximation of the geometric distortion of an image after digital data is encoded in it, and is used to compensate for this distortion to facilitate extracting embedded digital messages from the image. Due to the errors in the approximation, a signal confidence metric is determined and used to weight message symbol estimates extracted from the image.

First claim

1 . A method of determining a geometric transform of a reference signal in an image for extracting digital data, the method comprising:
with a programmed processor or digital logic circuit: transforming the image into a frequency domain to produce a discrete frequency domain representation of the image at integer coordinates; estimating phase at non-integer coordinates within the discrete frequency domain representation of the image according to a point spread function;
for plural reference signal components of a reference signal:
transforming coordinates of a reference signal component according to a candidate geometric transform;
updating the coordinates of the reference signal component to a location within a neighborhood of the discrete frequency domain representation around the coordinates of the transformed reference signal component, the location corresponding to a highest frequency magnitude of the frequency magnitudes within the neighborhood, wherein magnitude values at non-integer locations in the neighborhood are computed using the phase estimated at non-integer locations;
determining a new geometric transform that transforms the reference signal components to the updated coordinates; and applying the new geometric transform to extract encoded digital data from the image.

2 . The method of claim 1 wherein estimating phase at non-integer coordinates comprises estimating complex frequency components at the non-integer coordinates by applying weights obtained from a point spread function to neighboring complex frequency components at integer locations;
and wherein the frequency magnitudes within the neighborhood are magnitudes of the complex frequency components in the neighborhood.

3 . The method of claim 1 wherein the reference signal components comprise peaks in the frequency domain

4 . The method of claim 1 wherein the reference signal component comprises a sinusoid.

5 . The method of claim 1 including:
determining a signal confidence metric for a reference signal in an image block based on applying the new geometric transform to approximate geometric distortion of the image block; and
weighting digital message elements extracted from the image block by the signal confidence metric.

6 . The method of claim 1 comprising:
for geometric transform candidates, performing the transforming and the measuring of correlation with a lower frequency subset of the plural reference signal components to determine a subset of the geometric transform candidates for further refinement; and
in plural refinement stages, performing the transforming, the measuring of correlation, the updating of coordinates and the determining of a new geometric transform on the subset of the geometric transform candidates.

7 . The method of claim 1 comprising:
executing instructions on one or more processors to execute the acts of transforming, measuring, updating, determining the new geometric transform, and applying the new geometric transform.

8 . The method of claim 1 comprising:
performing the acts of transforming, measuring, updating, determining the new geometric transform, and applying the new geometric transform in special purpose digital logic circuitry.

9 . A non-transitory computer readable medium, on which is stored instructions, which when executed by a processor, perform a method of determining a geometric transform of a reference signal in an image for extracting digital data, the method comprising:
transforming the image into a frequency domain to produce a discrete frequency domain representation of the image at integer coordinates; estimating phase at non-integer coordinates within the discrete frequency domain representation of the image according to a point spread function;
for plural reference signal components of a reference signal:
transforming coordinates of a reference signal component according to a candidate geometric transform;
updating the coordinates of the reference signal component to a location within a neighborhood of the discrete frequency domain representation around the coordinates of the transformed reference signal component, the location corresponding to a highest frequency magnitude of the frequency magnitudes within the neighborhood, wherein magnitude values at non-integer locations in the neighborhood are computed using the phase estimated at non-integer locations;
determining a new geometric transform that transforms the reference signal components to the updated coordinates; and applying the new geometric transform to extract encoded digital data from the image.

10 . An apparatus for digital data extraction, the apparatus comprising;
an image sensor, memory for storing an image captured by the image sensor and software instructions; one or more processors coupled to the memory, the one or more processors configured to execute the software instructions to: apply a geometric transform candidate to a reference signal component to obtain transformed coordinates; determine new coordinates of the reference signal component within a neighborhood around the coordinates of the transformed reference signal component by locating a highest magnitude frequency component within the neighborhood, wherein magnitudes at non-integer locations in the neighborhood are obtained based on weighting of frequency components at integer locations according to a point spread function; compute a new geometric transform that fits the reference signal to the new coordinates; and extract encoded digital data from the image at locations determined with the new geometric transform.

11 . The method of claim 1 , comprising:
with a programmed processor or digital logic circuit:
for plural reference signal components of a reference signal,
transforming coordinates of a reference signal component according to a candidate geometric transform;
measuring correlation between the transformed reference signal component and the image, the correlation comprising a combination of complex components of the frequency domain transform of the image at neighboring integer coordinates around the coordinates of the transformed reference signal component, wherein the complex combination combines complex components according to a phase relationship at the neighboring coordinates; and
assessing the candidate geometric transform based on the correlation.

12 . The method of claim 11 wherein the phase relationship comprises matching phase at upper right and lower left neighbors, and matching phase at upper left and lower right, and phase at the upper right and upper left differ by 180 degrees.

13 - 15 . (canceled)

16 . The method of claim 11 comprising:
for geometric transform candidates, performing the transforming and the measuring of correlation with a lower frequency subset of the plural reference signal components to determine a subset of the geometric transform candidates to refine; and
in plural refinement stages, performing the transforming, the measuring of correlation, the updating of coordinates and the determining of a new geometric transform on the subset of the geometric transform candidates.

17 - 29 . (canceled)

30 . The apparatus of claim 10 , comprising:
memory in which is stored a reference signal and an image, which includes the reference signal; a programmed processor, the programmed processor executing instructions to: transform components of the reference signal by applying a geometric transform to the components, the geometric transform starting with transform candidate seeds, the transform candidate seeds comprising rotation angle, s

Assignees

• Digimarc Corp

Inventors

• Sharma Ravi K
• Lord John D
• Lyons Robert G
• Alattar Osama M
• Boles Jacob L

Classifications

• G06T2201/0052

  Embedding of the watermark in the frequency domain · CPC title

• G06T2201/0065

  Extraction of an embedded watermark; Reliable detection · CPC title

• G06T1/0064 Primary

  Geometric transfor invariant watermarking, e.g. affine transform invariant · CPC title

• G06T2201/0081

  whereby both original and watermarked images are required at decoder, e.g. destination-based, non-blind, non-oblivious · CPC title

• G06T3/0006

  Physics · mapped topic