Method of generating descriptor for interest point in image and apparatus implementing the same

Patent metadata

Field	Value
Publication number	US-2016180187-A1
Application number	US-201514757579-A
Country	US
Kind code	A1
Filing date	Dec 23, 2015
Priority date	Dec 23, 2014
Publication date	Jun 23, 2016
Grant date	—

Abstract

A method of generating a descriptor includes determining, in an image, a circular region having a center at an interest point to be described; extracting strips extending in the circular region in mutually different directions and intersecting together at the interest point; dividing each of the strips into patches; characterizing each of the patches using Haar features to acquire elements of the descriptor corresponding to the respective Haar features; and reordering the elements of the descriptor according to the strip corresponding to a maximum local.

First claim

1 . A method of generating a descriptor comprising the steps of:
determining, in an image, a circular region having a center at an interest point to be described; extracting strips extending in the circular region in mutually different directions and intersecting together at the interest point; dividing each of the strips into patches; characterize each of the patches using Haar features to acquire elements of the descriptor corresponding to the respective Haar features; and arranging the elements of the descriptor according to the strip corresponding to a maximum local.

2 . The method as claimed in claim 1 , wherein
the step of characterizing includes, for each of the patches, comparing mean intensities of pixels between two pixel groups, the pixels belonging to the two pixel groups being extracted from the patch according to each of the Haar features, to determine the elements of the descriptor corresponding to the respective Haar features.

3 . The method as claimed in claim 2 , wherein
the step of characterizing further includes, for each of the patches, determining, based on the comparing, which of the two pixel groups has the greater mean intensity, to determine the elements of the descriptor corresponding to the respective Haar features (.

4 . The method as claimed in claim 2 , wherein
the step of characterizing further includes, for each of the patches, calculating, based on the comparing, a difference between the respective mean intensities of the two pixel groups, to determine the elements of the descriptor corresponding to the respective Haar features.

5 . The method as claimed in claim 2 , wherein
the mean intensities are calculated by employing an integral image technique.

6 . The method as claimed in claim 1 , wherein
the step of arranging the elements of the descriptor includes:
determining one of the strips corresponding to a maximum local; and
arranging the elements of the descriptor according to the determined one of the strips.

7 . The method as claimed in claim 6 , wherein
the strip that corresponds to a maximum local is determined by determining which one of the strips has a maximum sum or variance of the intensities of the pixels included in the strip; and the step of arranging the elements of the descriptor according to the determined one of the strips includes placing the elements belonging to the determined one of the strips at a first place and placing the remaining elements in strip units in a clockwise order of the strips.

8 . The method as claimed in claim 1 , wherein
the step of extracting the strips includes extracting four strips, the step of dividing each of the strips includes dividing each of the strips into four patches, the step of characterizing each of the patches using Haar features includes characterizing each of the patches using eight of the Haar features, and the descriptor has 128 of the elements.

9 . Apparatus for generating a descriptor, the apparatus comprising:
a processor that is adapted to execute stored instructions; and a memory that stores instructions, the memory storing processor executable code that, when executed by the processor, is adapted to: determine, in an image, a circular region having a center at an interest point to be described; extract strips extending in the circular region in mutually different directions and intersecting together at the interest point; divide each of the strips into patches; characterize each of the patches using Haar features to acquire elements of the descriptor corresponding to the respective Haar features; and arrange the elements of the descriptor according to the strip corresponding to a maximum local.

10 . The apparatus as claimed in claim 9 , wherein
the characterizing includes, for each of the patches, comparing mean intensities of pixels between two pixel groups, the pixels belonging to the two pixel groups being extracted from the patch according to each of the Haar features, to determine the elements of the descriptor corresponding to the respective Haar features.

11 . The apparatus as claimed in claim 10 , wherein
the characterizing further includes, for each of the patches, determining, based on the comparing, which of the two pixel groups has the greater mean intensity, to determine the elements of the descriptor corresponding to the respective Haar features.

12 . The apparatus as claimed in claim 10 , wherein
the characterizing further includes, for each of the patches, calculating, based on the comparing, a difference between the respective mean intensities of the two pixel groups, to determine the elements of the descriptor corresponding to the respective Haar features.

13 . The apparatus as claimed in claims 9 , wherein
the extracting the strips includes extracting four strips, the dividing each of the strips includes dividing each of the strips into four patches, the characterizing each of the patches using Haar features includes characterizing each of the patches using eight of the Haar features, and the descriptor has 128 of the elements.

14 . Computer program product downloadable from a communication network and/or recorded on a medium readable by computer and/or executable by a processor, comprising program code instructions for implementing the steps of:
determining, in an image, a circular region having a center at an interest point to be described; extracting strips extending in the circular region in mutually different directions and intersecting together at the interest point; dividing each of the strips into patches; characterizing each of the patches using Haar features to acquire elements of the descriptor corresponding to the respective Haar features; and arranging the elements of the descriptor according to the strip corresponding to a maximum local.

15 . Non-transitory computer-readable medium comprising a computer program product recorded thereon and capable of being run by a processor, including program code instructions for implementing the steps of:
determining, in an image, a circular region having a center at an interest point to be described; extracting strips extending in the circular region in mutually different directions and intersecting together at the interest point; dividing each of the strips into patches; characterizing each of the patches using Haar features to acquire elements of the descriptor corresponding to the respective Haar features; and arranging the elements of the descriptor according to the strip corresponding to a maximum local.

Assignees

• Thomson Licensing

Inventors

• Lefebvre Frederic
• Demoulin Vincent

Classifications

• G06V10/462 Primary

  Salient features, e.g. scale invariant feature transforms [SIFT] · CPC title

• G06K9/4647

  Physics · mapped topic

• G06T7/2033

  Physics · mapped topic

• G06K9/56

  Physics · mapped topic

• G06K9/4614 Primary

  Physics · mapped topic