Image recognition device, image recognition method, and computer-readable recording medium having image recognition program recorded thereon

Patent metadata

Field	Value
Publication number	US-10861182-B2
Application number	US-201816211772-A
Country	US
Kind code	B2
Filing date	Dec 6, 2018
Priority date	Jan 30, 2018
Publication date	Dec 8, 2020
Grant date	Dec 8, 2020

Abstract

Image recognition device that combines first part candidates and second part candidates to generate a plurality of pairs; that calculates first estimate value for each of the plurality of pairs, this value indicating estimated possibility that these candidates are connected; that executes, for each of the plurality of pairs, correcting first estimate value by using second estimate value as a weight in a state in which a part candidate that is one among a pair of part candidates matching of which has already been performed and that differs from both first part candidate and second part candidate is connected to either first part candidate or second part candidate, second estimate value indicating estimated possibility that the pair of part candidates matching of which has already been performed are connected; and that performs matching of first part candidates and second part candidates using corrected first estimate values.

First claim

The invention claimed is:

1. An image recognition device comprising a hardware processor, wherein the hardware processor:
executes, for each of a plurality of parts, processing of detecting a plurality of part candidates serving as candidates of the part from an image in which a plurality of objects having the plurality of parts are captured;
for a first part candidate and a second part candidate that are the one and the other of the part candidates with respect to which matching is to be performed, combines a plurality of the first part candidates and a plurality of the second part candidates to generate a plurality of pairs each including the first part candidate and the second part candidate;
calculates a first estimate value for each of the plurality of pairs, the first estimate value indicating an estimated possibility that the first part candidate and the second part candidate are connected;
executes, for each of the plurality of pairs, processing of correcting the first estimate value by using a second estimate value as a weight when a part candidate that is one among a pair of part candidates matching of which has already been performed and that differs from both the first part candidate and the second part candidate is connected to either the first part candidate or the second part candidate, the second estimate value indicating an estimated possibility that the pair of part candidates matching of which has already been performed are connected; and
performs matching of the plurality of first part candidates and the plurality of second part candidates based on the corrected first estimate values for the respective ones of the plurality of pairs;
wherein the hardware processor sets a value yielded by multiplying the first estimate value and the second estimate value as the corrected first estimate value.

2. The image recognition device according to claim 1 further comprising
a first storage that stores in advance connection information indicating a connection relationship of the plurality of parts,
wherein the hardware processor uses the connection information stored in the first storage to specify, from among the part candidates detected by the hardware processor, part candidates of two types that have a relationship of the first part candidate and the second part candidate.

3. The image recognition device according to claim 1 further comprising
a second storage,
wherein, for each of the plurality of objects, the hardware processor executes processing of storing, to the second storage, a plurality of part candidates having been connected in one as a result of the matching as the plurality of parts that the single object has.

4. An image recognition method that:
executes, for each of a plurality of parts, processing of detecting a plurality of part candidates serving as candidates of the part from an image in which a plurality of objects having the plurality of parts are captured;
for a first part candidate and a second part candidate that are the one and the other of the part candidates with respect to which matching is to be performed, combines a plurality of the first part candidates and a plurality of the second part candidates to generate a plurality of pairs each including the first part candidate and the second part candidate;
calculates a first estimate value for each of the plurality of pairs, the first estimate value indicating an estimated possibility that the first part candidate and the second part candidate are connected;
executes, for each of the plurality of pairs, processing of correcting the first estimate value by using a second estimate value as a weight when a part candidate that is one among a pair of part candidates matching of which has already been performed and that differs from both the first part candidate and the second part candidate is connected to either the first part candidate or the second part candidate, the second estimate value indicating an estimated possibility that the pair of part candidates matching of which has already been performed are connected; and
performs matching of the plurality of first part candidates and the plurality of second part candidates based on the corrected first estimate values for the respective ones of the plurality of pairs;
wherein the hardware processor sets a value yielded by multiplying the first estimate value and the second estimate value as the corrected first estimate value.

5. A computer-readable recording medium having recorded thereon an image recognition program that causes a computer to:
execute, for each of a plurality of parts, processing of detecting a plurality of part
candidates serving as candidates of the part from an image in which a plurality of objects having the plurality of parts are captured;
for a first part candidate and a second part candidate that are the one and the other of the part candidates with respect to which matching is to be performed, combine a plurality of the first part candidates and a plurality of the second part candidates to generate a plurality of pairs each including the first part candidate and the second part candidate;
calculate a first estimate value for each of the plurality of pairs, the first estimate value indicating an estimated possibility that the first part candidate and the second part candidate are connected;
execute, for each of the plurality of pairs, processing of correcting the first estimate value by using a second estimate value as a weight when a part candidate that is one among a pair of part candidates matching of which has already been performed and that differs from both the first part candidate and the second part candidate is connected to either the first part candidate or the second part candidate, the second estimate value indicating an estimated possibility that the pair of part candidates matching of which has already been performed are connected; and
perform matching of the plurality of first part candidates and the plurality of second part candidates based on the corrected first estimate values for the respective ones of the plurality of pairs;
wherein the hardware processor sets a value yielded by multiplying the first estimate value and the second estimate value as the corrected first estimate value.

Assignees

• Konica Minolta Inc

Inventors

• Sekii Taiki

Classifications

• G06T7/77 Primary

  using statistical methods · CPC title

• G06V10/422

  for representing the structure of the pattern or shape of an object therefor · CPC title

• G06V40/103

  Static body considered as a whole, e.g. static pedestrian or occupant recognition · CPC title

• G06T7/70 Primary

  Determining position or orientation of objects or cameras (camera calibration G06T7/80) · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title