Registration of SAR images by mutual information

Patent metadata

Field	Value
Publication number	US-9495747-B2
Application number	US-201414785215-A
Country	US
Kind code	B2
Filing date	Apr 23, 2014
Priority date	Apr 24, 2013
Publication date	Nov 15, 2016
Grant date	Nov 15, 2016

Abstract

A method for registering an image using a similarity criterion based on mutual information. The image to be registered is compared with a plurality of reference representations of an object, each reference representation being of a plurality of homogeneous zones. The mutual information between the image to be registered and each reference representation is calculated, on the set of homogeneous zones. The registration is given by the reference representation corresponding to the highest mutual information. The method can be advantageously applied to an aircraft aided navigation by registering images obtained by a synthetic aperture radar.

First claim

The invention claimed is:

1. A method for registering an image of an object with respect to a plurality of reference representations of the object, each reference representation being of a set of homogeneous zones, each homogeneous zone having a homogeneous intensity level, the method comprising:
calculating, for each reference representation of the plurality, mutual information between the image and the reference representation, on the set of homogeneous zones of the reference representation;
comparing the mutual information thus calculated and selected the reference representation of the plurality for which the mutual information is highest.

2. The method for registering an image according to claim 1 , wherein, for each reference representation, the mutual information is determined by:
calculating entropy of the image on the set of homogeneous zones of the reference representation;
calculating, for each homogeneous zone, a weighting factor, associated with the zone, corresponding to the ratio between the area of the zone and total area of the set of the homogeneous zones;
calculating, for each homogeneous zone, the entropy of the image on the zone;
calculating the difference between the entropy of the image on the set of the homogeneous zones and the weighted sum of the entropies of the image on the different homogeneous zones, each zone being weighted by its weighting factor.

3. The method for registering an image according to claim 2 , wherein the calculating the entropy of the image on the set of the homogeneous zones comprises estimating pixel intensity probability density in the set of the homogeneous zones, by a histogram estimation or a core estimation.

4. The method for registering an image according to claim 2 , wherein the calculating the entropy of the image on each homogeneous zone comprises estimating pixel intensity probability density in the homogeneous zone, by a histogram estimation or a core estimation.

5. The method for registering an image according to claim 1 , wherein the reference representations are obtained by a segmentation into homogeneous zones of real images of the object, each real image having been taken by a sensor having predetermined position and attitude with respect to the object.

6. The method for registering an image according to claim 1 , wherein the reference representations are obtained from a three-dimensional digital model of the object, each reference representation being associated with a position and an attitude of the sensor with respect to the object.

7. The method for registering an image according to claim 6 , wherein the three-dimensional digital model of the object comprises homogeneous zones, each homogeneous zone of the digital model being defined by a contour, each contour being defined by a set of points and, optionally, by an interpolation function.

8. The method for registering an image according to claim 7 , wherein the contour of a homogeneous zone is defined by a polygon passing through the points of the set of points.

9. A synthetic aperture radar-aided aircraft navigation method, comprising:
obtaining at least one image of a zone overflown by the aircraft; and
registering the image with respect to a plurality of reference representations of the overflown zone, using the method for registering an image according to claim 1 , each reference representation being associated with a position and an attitude of the aircraft in a geo-referenced frame, and
wherein aircraft navigation information is deduced from position and attitude associated with the reference representation for which the mutual information is the highest.

10. The aided navigation method according to claim 9 , wherein the reference representations are obtained from a Field Digital Model stored in a database on-board the aircraft, the field digital model comprising a plurality of homogeneous zones, each homogeneous zone of the field digital model being defined by a contour, each contour being itself defined by a set of points of coordinates stored in the database and, optionally, by an interpolation function.

11. The aided navigation method according to claim 10 , wherein the homogeneous zones of the field digital model are classified by types, one type being associated with each of the zones and stored in relation with it in the database.

12. The aided navigation method according to claim 11 , wherein, for at least one reference representation, the homogeneous zones of the reference representation are obtained using a geometrical transformation of the homogeneous zones of the field digital model, the geometrical transformation depending on a position and an attitude of the aircraft.

Assignees

• Commissariat Energie Atomique

Inventors

• Minvielle-Larrousse Pierre

Classifications

• G06T7/35 Primary

  using statistical methods · CPC title

• G06T2207/20116

  Active contour; Active surface; Snakes · CPC title

• G06T7/44

  using image operators, e.g. filters, edge density metrics or local histograms · CPC title

• G06T2207/10032

  Satellite or aerial image; Remote sensing · CPC title

• G06T2207/20076

  Probabilistic image processing · CPC title