Positioning feature points of human face edge

Patent metadata

Field	Value
Publication number	US-9916494-B2
Application number	US-201615080125-A
Country	US
Kind code	B2
Filing date	Mar 24, 2016
Priority date	Mar 25, 2015
Publication date	Mar 13, 2018
Grant date	Mar 13, 2018

Abstract

An example method for positioning feature points of a human face edge including fitting a profile edge curve of a human face edge in a human face image according to the human face image; calculating by using a preset human face edge feature point calculation model to obtain feature points of the human face edge in the human face image; calculating by using a preset convergence algorithm to obtain feature information of a convergence feature point corresponding to each feature point; determining respectively whether a distance from each feature point to the profile edge curve is greater than a distance from a respective corresponding convergence feature point to the profile edge curve; and determining whether a magnitude of overall position change of all the feature points of the human face edge in the human face image before and after the above determination is less than a preset threshold.

First claim

What is claimed is:

1. A method comprising:
fitting a profile edge curve of a human face edge in a human face image according to the human face image;
determining feature points of the human face edge in the human face image by using a preset human face edge feature point calculation model;
acquiring feature information of a respective feature point;
determining feature information of a convergence feature point corresponding to the respective feature point according to the feature information of the respective feature point;
determining whether a distance from the respective feature point to the profile edge curve is greater than a distance from a corresponding respective convergence feature point to the profile edge curve according to the feature information of the respective feature point and the respective convergence feature point;
acquiring the feature information of the respective convergence feature point corresponding to the feature point to serve as the feature information of the feature point, in response to determining that the distance from the respective feature point to the profile edge curve is greater than the distance from a corresponding respective convergence feature point to the profile edge curve; and
outputting the feature points and the feature information of the respective feature point through an output interface for determining a shape of the human face.

2. The method of claim 1 , further comprising keeping the feature information of the feature point unchanged, in response to determining that the distance from the respective feature point to the profile edge curve is not greater than the distance from a corresponding respective convergence feature point to the profile edge curve.

3. The method of claim 1 , further comprising:
determining whether a magnitude of overall position change of all the feature points of the human face edge in the human face image before and after the above determination is less than a preset threshold; and
using position information of the group of feature points as qualified feature points of human face edge, in response to determining that the magnitude is less than the preset threshold.

4. The method of claim 3 , further comprising:
using a preset convergence algorithm to obtain feature information of a convergence feature point corresponding to each feature point according to the feature information of each feature point, in response to determining that the magnitude is not less than the preset threshold.

5. The method of claim 3 , wherein the determining whether the magnitude of overall position change of all the feature points of the human face edge in the human face image before and after the above determination is less than the preset threshold includes:
calculating, according to feature information of all the feature points of the human face edge in the human face image before and after the above determination, the magnitude of overall position change of all the feature points of the human face edge in the human face image before and after the above determination;
determining whether the magnitude of the position change is less than the preset threshold; and
determining that feature information of each feature point of the human face edge in the human face image is unchanged, in response to determining that the magnitude is less than the preset threshold.

6. The method of claim 5 , further comprising returning to the calculating by using the preset convergence algorithm to obtain feature information of the convergence feature point corresponding to each feature point according to the feature information of each feature point, in response to determining that the magnitude is not less than the preset threshold.

7. The method of claim 5 , wherein the calculating the magnitude of overall position change of all the feature points of the human face edge in the human face image before and after the above determination includes:
acquiring a vector formed by connecting all feature points of the human face edge in the human face image before the above determination;
acquiring a vector formed by connecting all feature points of the human face edge in the human face image after the above determination; and
calculating a Euclidean distance between the vector formed by connecting all feature points of the human face edge in the human face image before the above determination and the vector formed by connecting all feature points of the human face edge in the human face image after the above determination to serve as the magnitude of overall position change of all the feature points of the human face edge in the human face image before and after the above determination.

8. The method of claim 1 , wherein the preset convergence algorithm includes:
Δ X 1 =R 0 *φ 0 +b 0 ; and

X k =X k−1 +R k−1 *φ k−1 +b k−1 ,

wherein:
X k is a vector formed by connecting convergence feature points obtained after calculation by using the convergence algorithm for the k th time, and X k−1 is a vector formed by connecting convergence feature points obtained after calculation by using the convergence algorithm for the (k−1) th time;
ΔX 1 is an offset value between a vector X 1 formed by connecting convergence feature points obtained after calculation by using the convergence algorithm for the first time and a vector X 0 formed by connecting the feature points of the human face edge in the human face image; φ 0 is a vector formed by connecting gradient direction features of the feature points before calculation by using the convergence algorithm for the first time; and R0 and b0 are preset coefficients; and
R k−1 and b k−1 are coefficients after calculation by using the convergence algorithm for the (k−1) th time, and φ k−1 is a vector formed by connecting gradient direction features of convergence feature points obtained after calculation by using the convergence algorithm for the (k−1) th time.

9. The method of claim 1 , further comprising:
prior to fitting the profile edge curve of human face edge in the human face image according to the human face image,
acquiring the human face image of the human face;
separating skin color regions and non-skin color regions in the human face image by using a preset skin color model; and
filtering out the non-skin color regions in the human face image.

10. The method of claim 9 , further comprising:
after acquiring the human face image of the human face,
expanding or clipping the acquired human face image into a human face image having a set size or format.

11. The method of claim 8 , wherein the preset skin color model includes a Gaussian mixture model.

12. The method of claim 1 , wherein the fitting the profile edge curve of human face edge in the human face image according to the human face image includes:
detecting strong edge of the human face in the human face image; and
fitting the profile edge curve of human face edge in the human face image according to the detected strong edge of the human face in the human face image.

13. The method of claim 12 , further comprising acquiring the strong edge of the human face in the human face image by detecting using a Canny edge detection operator.

14. The method of claim 12 , wherein the fitting the profile edge curve of human face edge in the human face image includes:
filtering the detected strong edge of the human face in the human face image to obtain the profile curve of the human face in the human face image;
making at least one ray according to a preset angle by using at least one feature point corresponding to the five sense organs of th

Assignees

• Alibaba Group Holding Ltd

Inventors

• Zou Jianfa
• Huang Ying
• Chen Yan

Classifications

• G06V40/171

  Local features and components; Facial parts (eye characteristics G06V40/18); Occluding parts, e.g. glasses; Geometrical relationships · CPC title

• G06T7/12 Primary

  Edge-based segmentation · CPC title

• G06T2207/30201

  Face · CPC title

• G06T7/149

  involving deformable models, e.g. active contour models · CPC title

• G06V10/56

  relating to colour · CPC title