Feature point estimation device, feature point position estimation system, feature point position estimation method, and feature point position estimation program

Patent metadata

Field	Value
Publication number	US-2020327368-A1
Application number	US-201716305909-A
Country	US
Kind code	A1
Filing date	May 22, 2017
Priority date	Jun 2, 2016
Publication date	Oct 15, 2020
Grant date	—

Abstract

A feature point position estimation device is provided. The feature point position estimation device includes a subject detection section for detecting a subject region from a subject image, a feature point positioning section for positioning a feature point at a preliminarily prepared initial feature point position with respect to the subject region, a feature amount acquisition unit for acquiring a feature amount of the feature points arranged, a regression calculation unit for calculating a deviation amount of a position of a true feature point with respect to the position of the feature point by performing a regression calculation on the feature amount, and a repositioning unit for repositioning the feature points based on the deviation amount. The regression calculation unit calculates the deviation amount by converting the feature amount in a matrix-resolved regression matrix.

First claim

What is claimed is:

1 . A feature point position estimation device, comprising:
a subject detection unit configured to detect a subject area from a subject image; a feature point positioning unit configured to position a feature point at an initial feature point position prepared in advance for the subject region; a feature amount acquisition unit configured to acquire a feature amount of the arranged feature point; a regression calculation unit configured to calculate a deviation amount of a position of a true feature point with respect to the position of the feature point by performing a regression calculation on the feature amount; and a repositioning unit configured to reposition the feature points based on the deviation amount; wherein the feature amount acquisition unit acquires a feature point arranged in the feature point positioning unit and a feature amount of a feature point repositioned in the repositioning unit, the regression calculation unit calculates the deviation amount by converting the feature amount in a matrix-resolved regression matrix, and the repositioning unit repeats a plurality of times acquisition of the feature amount by the feature amount acquisition unit, calculation of the deviation amount by the regression calculation unit, and reposition of the feature points, and outputs a position of the repositioned feature point.

2 . The feature point position estimation device according to claim 1 , wherein
the regression matrix is decomposed into a basis matrix which is a real number matrix and a coefficient matrix which is a real number matrix.

3 . The feature point position estimation device according to claim 2 , wherein
the regression matrix is decomposed into the basis matrix and the coefficient matrix by singular value decomposition.

4 . The feature point position estimation device according to claim 1 , wherein
the regression matrix is decomposed into a basis matrix which is an integer matrix and a coefficient matrix which is a real number matrix.

5 . The feature point position estimation device according to claim 4 , wherein
the basis matrix is a binary matrix or a ternary matrix.

6 . The feature point position estimation device according to claim 4 , wherein
in the regression matrix, each column vector is individually decomposed.

7 . The feature point position estimation device according to claim 4 , wherein
the regression matrices are collectively and matrix-decomposed.

8 . The feature point position estimation device according to claim 1 , wherein
the feature point position estimation device estimates a position of a feature point with respect to subject images of a plurality of consecutive frames, and in a process of repeating acquisition of the feature amount by the feature amount acquisition unit, calculation of the deviation amount by the regression calculation unit, and reposition of the feature points in the previous frame a plurality of times, the feature point positioning unit arranges feature points so that the positions of the repositioned feature points are located as the initial feature point position prepared in advance of the current frame.

9 . The feature point position estimation device according to claim 1 , further comprising:
an evaluation unit configured to obtain a score of the feature point by linearly transforming the feature amount of the feature point rearranged in the repositioning unit.

10 . The feature point position estimation device according to claim 9 , wherein
the evaluation unit groups the plurality of feature points repositioned by the repositioning unit, and obtains the score for each group.

11 . The feature point position estimation device according to claim 1 , wherein
the regression calculation unit performs the regression calculation only for some of the feature points with a high priority among the plurality of feature points, repeats the acquisition of the feature amount by the feature amount acquisition unit, the calculation of the degree of deviation by the regression calculation unit, and the reposition of the feature point a plurality of times, increases the number of feature points for which the regression calculation is performed according to the priority, and performs the regression calculation for all feature points.

12 . The feature point position estimation device according to claim 1 , wherein
the subject detection unit detects the subject area by extracting feature amount from the plurality of blocks of the subject image and performing identification processing, and the feature amount acquisition unit acquires the feature amount extracted by the subject detection unit as a feature amount of the feature point.

13 . The feature point position estimation device according to claim 12 , wherein
the feature amount acquisition unit acquires a feature amount of the block to which the feature point belongs as a feature amount of the feature point.

14 . A feature point position estimation system, comprising:
a subject detection unit configured to detect a subject area from a subject image; a feature point positioning unit configured to position a feature point at an initial feature point position prepared in advance for the subject region; a feature amount acquisition unit configured to acquire a feature amount of the arranged feature point; a regression calculation unit configured to calculate a deviation amount of a position of a true feature point with respect to the position of the feature point by performing a regression calculation on the feature amount; and a repositioning unit configured to reposition the feature points based on the deviation amount; wherein the feature amount acquisition unit acquires a feature point arranged in the feature point positioning unit and a feature amount of a feature point repositioned in the repositioning unit, the regression calculation unit calculates the deviation amount by converting the feature amount in a matrix-resolved regression matrix, and the repositioning unit repeats a plurality of times acquisition of the feature amount by the feature amount acquisition unit, calculation of the deviation amount by the regression calculation unit, and reposition of the feature points, and outputs a position of the repositioned feature point.

15 . A feature point position estimation program that is executed by a computer and causes the computer to function as, when executed on a computer,
a subject detection unit configured to detect a subject area from a subject image; a feature point positioning unit configured to position a feature point at an initial feature point position prepared in advance for the subject region; a feature amount acquisition unit configured to acquire a feature amount of the arranged feature point; a regression calculation unit configured to calculate a deviation amount of a position of a true feature point with respect to the position of the feature point by performing a regression calculation on the feature amount; and a repositioning unit configured to reposition the feature points based on the deviation amount; wherein the feature amount acquisition unit acquires a feature point arranged in the feature point positioning unit and a feature amount of a feature point repositioned in the repositioning unit, the regression calculation unit calculates the deviation amount by converting the feature amount in a matrix-resolved regression matrix, and the repositioning unit repeats a plurality of times acquisition of the feature amount by the feature amount acquisition unit, calculation of the deviatio

Assignees

• Denso Corp

Inventors

• Ambai Mitsuru
• Munaoka Yutaka
• Omi Takuhiro

Classifications

• G06T7/00 Primary

  Image analysis · CPC title

• G06V40/168

  Feature extraction; Face representation · CPC title

• G06V10/462

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

• G06T7/73

  using feature-based methods · CPC title

• G06F17/16

  Matrix or vector computation {, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization (matrix transposition G06F7/78)} · CPC title