Methods and systems for generating composite image descriptors

Patent metadata

Field	Value
Publication number	US-11461578-B2
Application number	US-202117167846-A
Country	US
Kind code	B2
Filing date	Feb 4, 2021
Priority date	Feb 4, 2021
Publication date	Oct 4, 2022
Grant date	Oct 4, 2022

Abstract

An illustrative image descriptor generation system generates a descriptor listing that includes a plurality of image descriptors corresponding to different feature points included within an image. Based on the descriptor listing, the system generates a geometric map representing the plurality of image descriptors in accordance with respective geometric positions of the corresponding feature points of the image descriptors within the image. Based on the geometric map, the system determines a proximity listing for a primary image descriptor within the plurality of image descriptors. The proximity listing indicates a subset of image descriptors that are geometrically proximate to the primary image descriptor within the image. Based on the proximity listing, the system selects a secondary image descriptor from the subset of image descriptors and combines the primary and secondary image descriptors to form a composite image descriptor. Corresponding methods and systems are also disclosed.

First claim

What is claimed is:

1. A method comprising:
generating, by an image descriptor generation system based on an image, a descriptor listing that includes a plurality of image descriptors corresponding to different feature points included within the image;
generating, by the image descriptor generation system based on the descriptor listing, a geometric map representing the plurality of image descriptors in accordance with respective geometric positions of the corresponding feature points of the plurality of image descriptors within the image;
determining, by the image descriptor generation system based on the geometric map, a proximity listing for a primary image descriptor within the plurality of image descriptors, the proximity listing indicating a subset of image descriptors, from the plurality of image descriptors, that are geometrically proximate to the primary image descriptor within the image;
selecting, by the image descriptor generation system based on the proximity listing, a secondary image descriptor from the subset of image descriptors; and
combining, by the image descriptor generation system, the primary image descriptor and the secondary image descriptor to form a composite image descriptor.

2. The method of claim 1 , wherein:
the image depicts a real-world scene and is captured and provided to the image descriptor generation system in real time by an extended reality generation system; and
the method further comprises providing, by the image descriptor generation system, the composite image descriptor to the extended reality generation system for use by the extended reality generation system in identifying one or more predetermined target objects within the image.

3. The method of claim 1 , wherein the generating of the descriptor listing includes:
accessing the image in real time as the image is generated;
analyzing the image in real time to identify the different feature points included within the image; and
generating, in real time for each of the different feature points that are identified, a respective image descriptor to be among the plurality of image descriptors included in the descriptor listing.

4. The method of claim 1 , wherein:
the plurality of image descriptors included in the descriptor listing includes:
a first group of image descriptors generated for the image at a first image resolution, and
a second group of image descriptors generated for the image at a second image resolution that is less than the first image resolution; and

the generating of the geometric map, the determining of the proximity listing, the selecting of the secondary image descriptor, and the combining of the primary and secondary image descriptors to form the composite image descriptor are each performed in a first stage for the first group of image descriptors, and in a second stage for the second group of image descriptors.

5. The method of claim 1 , wherein:
the generating of the geometric map includes:
partitioning the image into a plurality of bins arranged in horizontal rows and vertical columns, and
for each image descriptor of the plurality of image descriptors, assigning the image descriptor to a respective bin of the plurality of bins based on a position within the image of a respective feature point corresponding to the image descriptor; and

the determining of the proximity listing includes, for each bin of the plurality of bins that is within a search radius of a first bin to which the primary image descriptor is assigned:
determining whether the bin is assigned a first image descriptor of the plurality of image descriptors, and
if the bin is assigned the first image descriptor,

including the first image descriptor in the subset of image descriptors indicated by the proximity listing.

6. The method of claim 1 , wherein the selecting of the secondary image descriptor is based on a geometric distance between the primary image descriptor and the secondary image descriptor on the geometric map.

7. The method of claim 6 , wherein the selecting based on the geometric distance includes filtering the subset of image descriptors indicated by the proximity listing to exclude image descriptors that are more than a threshold distance away from the primary image descriptor on the geometric map.

8. The method of claim 1 , wherein the selecting of the secondary image descriptor is based on a similarity between the primary image descriptor and the secondary image descriptor.

9. The method of claim 8 , wherein the selecting based on the similarity includes filtering the subset of image descriptors indicated by the proximity listing to exclude image descriptors that are more similar to the primary image descriptor than a similarity threshold.

10. The method of claim 1 , wherein the selecting of the secondary image descriptor includes:
filtering the subset of image descriptors indicated by the proximity listing to exclude image descriptors that are:
more than a threshold distance away from the primary image descriptor on the geometric map, or
more similar to the primary image descriptor than a similarity threshold;

sorting the filtered subset of image descriptors from a least similar image descriptor to a most similar image descriptor; and
designating, using the filtered and sorted subset of image descriptors, the least similar image descriptor as the selected secondary image descriptor.

11. The method of claim 1 , further comprising:
selecting, by the image descriptor generation system and based on the proximity listing, one or more additional image descriptors from the subset of image descriptors; and
combining, by the image descriptor generation system, the primary image descriptor with each of the one or more additional image descriptors to form one or more additional composite image descriptors.

12. A system comprising:
a memory storing instructions; and
a processor communicatively coupled to the memory and configured to execute the instructions to:
generate, based on an image, a descriptor listing that includes a plurality of image descriptors corresponding to different feature points included within the image;
generate, based on the descriptor listing, a geometric map representing the plurality of image descriptors in accordance with respective geometric positions of the corresponding feature points of the plurality of image descriptors within the image;
determine, based on the geometric map, a proximity listing for a primary image descriptor within the plurality of image descriptors, the proximity listing indicating a subset of image descriptors, from the plurality of image descriptors, that are geometrically proximate to the primary image descriptor within the image;
select, based on the proximity listing, a secondary image descriptor from the subset of image descriptors; and
combine the primary image descriptor and the secondary image descriptor to form a composite image descriptor.

13. The system of claim 12 , wherein:
the image depicts a real-world scene and is captured and provided to the system in real time by an extended reality generation system; and
the processor is further configured to execute the instructions to provide the composite image descriptor to the extended reality generation system for use by the extended reality generation system in identifying one or more predetermined target objects within the image.

14. The system of claim 12 , wherein:
the plurality of image descriptors included in the descriptor listing includes:
a first group of image descriptors generated for the image at a first image resolution, and
a second group of image descriptors generated

Assignees

• Verizon Patent & Licensing Inc

Inventors

• Virodov Sergey
• Polavarapu Vinaya Kumar

Classifications

• G06F18/22 Primary

  Matching criteria, e.g. proximity measures · CPC title

• G06F18/253

  of extracted features · CPC title

• G06F18/2113

  by ranking or filtering the set of features, e.g. using a measure of variance or of feature cross-correlation · CPC title

• G06F18/211

  Selection of the most significant subset of features · CPC title

• G06V10/464 Primary

  using a plurality of salient features, e.g. bag-of-words [BoW] representations · CPC title