Skin detection method and electronic device

Patent metadata

Field	Value
Publication number	US-11798162-B2
Application number	US-201817260875-A
Country	US
Kind code	B2
Filing date	Sep 18, 2018
Priority date	Jul 16, 2018
Publication date	Oct 24, 2023
Grant date	Oct 24, 2023

Abstract

A skin detection method includes: dividing a region of interest in a face image into a highlighted region and a non-highlighted region; separately determining a first segmentation threshold of the highlighted region and a second segmentation threshold of the non-highlighted region; obtaining a binary image of the highlighted region based on the first segmentation threshold, and obtaining a binary image of the non-highlighted region based on the second segmentation threshold; fusing the binary image of the highlighted region and the binary image of the non-highlighted region; and identifying, based on a fused image, pores and/or blackheads included in the region of interest.

First claim

What is claimed is:

1. A skin detection method comprising:
obtaining a region of interest in a face image;
dividing the region of interest into a highlighted region and a non-highlighted region;
determining a first segmentation threshold of the highlighted region;
determining a second segmentation threshold of the non-highlighted region;
obtaining a first binary image of the highlighted region based on the first segmentation threshold;
obtaining a second binary image of the non-highlighted region based on the second segmentation threshold;
fusing the first binary image and the second binary image to obtain a fused image; and
identifying, based on the fused image, pores, blackheads, or both pores and blackheads in the region of interest.

2. The skin detection method of claim 1 , further comprising:
performing first graying processing on the region of interest to obtain a first grayscale image;
performing binarization processing on the first grayscale image to obtain a third binary image;
performing connected component analysis on the third binary image to obtain at least one first-connected component; and
filtering out a first connected component, of the at least one connected component, that comprises an area less than a first preset threshold to obtain the highlighted region, wherein a region other than the highlighted region in the region of interest is the non-highlighted region.

3. The skin detection method of claim 2 , further comprising:
performing grayscale transformation on the region of interest to obtain a second grayscale image;
multiplying each pixel of a plurality of pixels in the region of interest by a respective weight value, of a plurality of weight values, corresponding to the pixel to obtain a multiplied image, wherein each weight value of the plurality of weight values based on a grayscale value of a respective pixel in the second grayscale image;
performing grayscale transformation on the multiplied image to obtain a third grayscale image;
performing min-max normalization on the third grayscale image to obtain a min-max normalized third grayscale image; and
performing grayscale inversion processing on the min-max normalized third grayscale image to obtain the first grayscale image.

4. The skin detection method of claim 2 , further comprising:
performing expansion processing, corrosion processing, or both expansion processing and corrosion processing on the third binary image to obtain a processed image; and
obtaining at least one second connected component in the processed image.

5. The skin detection method of claim 1 , further comprising:
performing first graying processing on the region of interest to obtain a first grayscale image;
obtaining a plurality of minimum grayscale values of pixels in the highlighted region in the first grayscale image and a plurality of neighboring pixels of each of the pixels, wherein a quantity of neighboring pixels is equal to a multiple of 4; and
using an average value of the minimum grayscale values as the first segmentation threshold.

6. The skin detection method of claim 1 , further comprising:
performing graying processing on the region of interest to obtain a grayscale image;
obtaining a plurality of minimum grayscale values of pixels in the non-highlighted region in the grayscale image and a plurality of neighboring pixels of each of the pixels; and
using an average value of the minimum grayscale values as the second segmentation threshold.

7. The skin detection method of claim 1 , further comprising:
performing corrosion processing, expansion processing, or both corrosion processing and expansion processing on the fused image to obtain a processed image; and
identifying, based on the processed image the pores, the blackheads, or both the pores and the blackheads in the region of interest.

8. The skin detection method of claim 1 , further comprising:
removing a nose shadow region from the fused image to obtain a fourth image; and
determining, based on the fourth image, the pores, the blackheads, or both the pores and the blackheads, in the region of interest.

9. The skin detection method of claim 8 , further comprising:
performing graying processing on the region of interest to obtain a grayscale image;
performing binarization processing on the grayscale image to obtain a third binary image;
performing connected component analysis on the third binary image to obtain at least one first connected component; and
filtering out a connected component, of the at least one first connected component, comprising an area less than a second preset threshold to obtain the nose shadow region.

10. The skin detection method of claim 9 , further comprising:
performing expansion processing, corrosion processing, or both expansion processing and corrosion processing on the third binary image to obtain a processed binary image; and
obtaining at least one second connected component in the processed binary image.

11. The skin detection method of claim 5 , further comprising:
performing grayscale transformation on the region of interest to obtain a second grayscale image;
performing min-max normalization on the second grayscale image to obtain a min-max normalized second grayscale image;
performing Gaussian differential filtering on the min-max normalized second grayscale image to obtain a filtered image;
performing min-max normalization on the filtered image to obtain a min-max normalized filtered image;
obtaining a histogram of the min-max normalized filtered image; and
based on the histogram, setting a first grayscale value of a first type of pixels in the filtered image to 0 and setting a second grayscale value of a second type of pixels in the filtered image to 255, wherein the first grayscale value is less than or equal to a minimum grayscale value of other pixels in the filtered image, wherein a first percentage of a first quantity of the first type of pixels in a total quantity of pixels in the filtered image is a %, wherein the second grayscale value is greater than or equal to a maximum grayscale value of other pixels in the filtered image, wherein a second percentage of a second quantity of the second type of pixels in the total quantity is b %, and wherein a and b are both positive integers.

12. The skin detection method of claim 1 , further comprising:
obtaining at least one connected component comprised in the fused image;
obtaining a parameter for each connected component of the at least one connected component, wherein the parameter comprises at least one of a radius, a degree of circularity, or an internal-external color variation value; and
identifying, for each connected component of the at least one connected component based on the parameter for the connected component, whether a first location corresponding to the connected component is a pore or a blackhead, wherein the degree of circularity represents a degree of similarity between a shape of the connected component and a circle, and wherein the internal-external color variation value is a difference or a ratio between an internal pixel value and an external pixel value corresponding to the connected component at a second location of the region of interest.

13. The skin detection method of claim 1 , further comprising:
processing an image corresponding to identified pores to obtain a quantized pore indicator, wherein the quantized pore indicator comprises at least one of a pore area ratio, a pore density, an average pore internal-external color variation value, or a ratio of a pore internal-external color variation value-weighted area, wherein the pore area ratio is a first rat

Assignees

• Honor Device Co Ltd

Inventors

• Ding Xin
• Dong Chen
• Hu Hongwei
• Jiang Yongtao
• Gao Wenmei

Classifications

• G06T7/0012 Primary

  Biomedical image inspection · CPC title

• G06F18/25

  Fusion techniques · CPC title

• G06V10/28

  Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns · CPC title

• G06V10/48

  by mapping characteristic values of the pattern into a parameter space, e.g. Hough transformation · CPC title

• G06V10/50

  by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis · CPC title