CN111738984B - Skin image spot evaluation method and system based on watershed and seed filling - Google Patents
Skin image spot evaluation method and system based on watershed and seed filling Download PDFInfo
- Publication number
- CN111738984B CN111738984B CN202010473389.7A CN202010473389A CN111738984B CN 111738984 B CN111738984 B CN 111738984B CN 202010473389 A CN202010473389 A CN 202010473389A CN 111738984 B CN111738984 B CN 111738984B
- Authority
- CN
- China
- Prior art keywords
- image
- skin
- spot
- gray
- pixel
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000011156 evaluation Methods 0.000 title claims description 15
- 238000000034 method Methods 0.000 claims abstract description 57
- 238000001514 detection method Methods 0.000 claims abstract description 17
- 239000011148 porous material Substances 0.000 claims description 28
- 238000012545 processing Methods 0.000 claims description 26
- 239000004576 sand Substances 0.000 claims description 26
- 238000004364 calculation method Methods 0.000 claims description 24
- 210000004209 hair Anatomy 0.000 claims description 24
- 238000005286 illumination Methods 0.000 claims description 15
- 239000011159 matrix material Substances 0.000 claims description 14
- 238000005260 corrosion Methods 0.000 claims description 13
- 230000007797 corrosion Effects 0.000 claims description 13
- 238000003384 imaging method Methods 0.000 claims description 9
- 239000003086 colorant Substances 0.000 claims 2
- 230000003628 erosive effect Effects 0.000 claims 1
- 238000012946 outsourcing Methods 0.000 claims 1
- 238000011158 quantitative evaluation Methods 0.000 abstract description 15
- 238000010586 diagram Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 2
- 230000006798 recombination Effects 0.000 description 2
- 238000005215 recombination Methods 0.000 description 2
- 230000003712 anti-aging effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000004451 qualitative analysis Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 230000009759 skin aging Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration using local operators
- G06T5/30—Erosion or dilatation, e.g. thinning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20092—Interactive image processing based on input by user
- G06T2207/20101—Interactive definition of point of interest, landmark or seed
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30088—Skin; Dermal
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Quality & Reliability (AREA)
- Radiology & Medical Imaging (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Geometry (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a quantitative evaluation method and a quantitative evaluation system for skin image spots based on watershed and seed filling algorithm, wherein the system comprises skin image acquisition hardware equipment, a computer server side and a mobile phone client side; according to the pixel color value information of the RGB color space of the skin image, calculating attribute characteristic values such as the size, the distance, the number, the darkness, the duty ratio and the like of a skin image spot, quantitatively calculating a plurality of attribute values of the skin image spot, and evaluating the attribute of the skin spot. The spot detection method has the advantages of high spot detection accuracy, high speed and good market application prospect and value.
Description
Technical Field
The invention relates to a skin spot detection and evaluation technology, in particular to a skin image spot quantitative evaluation method and system based on watershed and seed filling algorithm, and belongs to the technical field of computer graphics and skin image processing application.
Background
The states of various attributes of the skin surface are important indexes of skin aging and anti-aging research, and are also one of important indexes for objectively evaluating skin care products. The objective and quantitative detection and calculation of skin images is one of research hotspots in the field of skin image processing, and one important application is the quantitative evaluation and calculation of skin surface spots.
At present, the accuracy of detecting the surface state of the skin at home and abroad and calculating the attribute indexes is not very high, and the method mainly comprises detection research and application in two directions of machine learning and computer images. With the rapid development of computer image processing technology, people begin to discuss the use of digital image processing technology to extract the features of skin surface, and the traditional qualitative analysis of skin surface condition is improved to accurately and quantitatively calculate the skin surface attribute, so as to improve the accuracy of detection and evaluation.
Features of skin images include color, texture, pores, glossiness, spots, fat spots, etc., where the size, spacing, number, darkness, duty cycle, etc. of the spots are important metrics of the skin. In recent years, a computer image method for detecting spots is proposed, and the main idea is to perform simple threshold segmentation on the color value of each pixel in an RGB (or HSV) color space to obtain spot pixels, wherein the method has more pseudo spot pixels and affects the quantitative evaluation accuracy; only the pixel point is studied, and the line and surface concepts are not available, so that many indexes such as the number of spots, the spot spacing and the like cannot be calculated.
Disclosure of Invention
The invention aims to realize a quantitative evaluation method of skin image spots based on watershed and seed filling algorithm, which is used for calculating attribute characteristic values such as the size, the distance, the number, the darkness, the duty ratio and the like of a skin image spot according to pixel color value information of RGB color space of the skin image, quantitatively calculating a plurality of attribute values of the skin image spot, and realizing the evaluation of the attribute of the skin spot.
In the invention, the skin image comes from a digital image acquisition device, and the macro skin image acquisition device acquires the image to obtain skin image data. The algorithm of the invention uses the collected skin image as a unique data source to calculate quantitative values of a plurality of attributes of skin spots; acquiring a gray image through fixed combination of three different color components; removing hair and white noise by a threshold method; the mathematical morphological closing operation fills in the sand hole noise in the speckle foreground, and the opening operation eliminates pores on the background; obtaining a continuous closed dividing line of each spot block based on a watershed algorithm of the computer image; a seed filling algorithm based on computer graphics fills the inside of the closed dividing line of the spot; calculating the area, perimeter, roundness, central point and darkness of a single spot; the average diameter of the spots, the average distance between the spots, the number of spots, the average darkness and the spot duty ratio of the whole image are counted.
The skin image spot quantitative evaluation method based on the watershed and the seed filling algorithm can process the skin micro-distance digital image with the same resolution, calculate and obtain a plurality of attribute index quantitative values for measuring a skin image spot, and the values can identify spot characteristics of the skin surface of the micro-distance image, and mainly comprise the following steps:
1) Graying the skin image by utilizing three different color components in a skin image color space (RGB) to obtain a skin gray image;
2) Removing hairs and white noise in the image from the skin gray image, and unifying the illumination intensity of the image;
3) Further processing the skin gray image obtained in the step 2) to remove pores and sand hole noise;
4) Processing the skin gray image obtained in the step 3) by using a computer image watershed algorithm to obtain a continuous closed parting line (contour line) of skin spots in the image;
5) Filling the interior of a continuous closed dividing line of skin spots in the image by using a computer graphics seed filling algorithm;
6) Calculating the area, perimeter, roundness, center point and darkness of a single spot of the skin in the image;
7) And counting to obtain the size, the interval, the number, the darkness and the duty ratio of the skin spots in the whole image.
Specifically, the method of the present invention comprises the steps of:
A. the recombination of three different color components of the skin image yields a gray image img1, which is specifically as follows:
A1. reading a skin image file into a memory;
the only parameter of the algorithm is an RGB space image file stored on a hard disk, and the file is read into a memory;
A2. graying the color image to obtain a gray image img1;
the three different color components of the RGB image are weighted and averaged to obtain a gray image img1, and the weighted average calculation method is as follows: g=r×0.7+g×0.1+b×0.2, where G is a gray value of a pixel, R, G, B is red, green, and blue color components of the pixel, and a gray value G value range is [0,255];
B. on the gray image img1, hair and white noise are removed, and the illumination intensity is unified, and the specific steps are as follows:
B1. calculating a gray image img1 gray average avg, wherein avg=the sum of all pixel gray values and dividing the sum by the number of pixels; will gray scale
The average avg is used as the background color;
B2. removing hair;
hair pixels are represented on a gray image img1 as small gray values, are almost black in color, a hair threshold v1 (such as v1=40+avg-127) is given, and all pixel points with gray values smaller than the threshold on the image img1 are reassigned to background color avg;
B3. white noise is removed;
white noise appears as a very large gray value in the gray image img1, the color is almost white, a white noise threshold v2 is given (v 1< v2, such as v2=220+avg-127), and all pixel points with gray values larger than the threshold are reassigned to be background color avg on the image img1;
B4. unifying the illumination intensity;
the uniform illumination intensity is to avoid calculation errors (comparing different pictures) caused by illumination differences of a plurality of skin images, and after hair and white noise are removed, the pixel gray value interval of the gray image img1 is [ v1, v2], the pixels of the gray image img1 are stretched to interval [0,255] in proportion, and the calculation method of the proportional stretching is as follows:
G1＝(G－v1)×255/(v2－v1)，
wherein G1 is the gray value of each pixel in the gray image after stretching; g is the gray value of each pixel in img1, and v1 and v2 are the hair threshold and the white noise threshold respectively;
the average gray value avg is avg1 (average gray value after stretching), avg1= (avg-v 1) ×255/(v 2-v 1);
after the gray image img1 is stretched, the gray value range of all pixels is [0,255];
C. pore and sand hole noise are removed from the gray image img1, and the specific contents are as follows:
C1. binarizing the gray level image to obtain a binarized image img2;
determining a binarization threshold Vt, wherein Vt < avg1-C, C is a constant (e.g. vt=avg1-30), and obtaining a value b of each pixel point on the binarized image img2, wherein the value b is determined according to a gray value v of the corresponding point of the pixel point on the gray image img1, which is specifically as follows:
c1.1 if v > Vt, b=0, identifying the pixel as a moire noise point in the background or speckle foreground;
c1.2 if v < = Vt, b = 1, identifying the pixel as a pore in the foreground or background of the blob.
C2. Removing pores in an image img2 background based on a mathematical morphology open operation method;
c2.1 sets a convolution Kernel1 for image processing; the convolution kernel is a weight definition function adopted in a pixel of an output image obtained by carrying out weighted average on the pixel in a small area in the input image; kernel1 is a square pixel matrix, wherein the center point of the square is a convolution Kernel origin, the convolution Kernel pixel value is 1 or 0, the pixel value of an inscribed circle part of the square is 1, the pixel value outside the inscribed circle and in the square is 0;
specifically, a convolution Kernel Kernel1 for image processing is set based on a mathematical morphology open operation method; the convolution Kernel is used for performing image processing on an input image, and performing weighted average on pixels in a small area in the input image to obtain each pixel of an output image, wherein a weight is defined by the convolution Kernel (namely a weight definition function), and Kernel1 is a square pixel square matrix a with a size edge2 x edge2 (edge 2 is an odd number, such as edge 2=29); in the square pixel matrix A, the center point of the square is a convolution kernel origin, the convolution kernel pixel value is 1 or 0, the pixel value of an inscribed circle part of the square is 1, the pixel value outside the inscribed circle is 0; the convolution kernel has 665 1 values and 176 0 values (the value of square pixel matrix A, when the value of edge2 is 29);
c2.2, regarding each pixel point p on the binarized image img2, setting the position of the pixel point p as a convolution Kernel origin, performing image corrosion operation by using a convolution Kernel Kernel1, and performing image expansion operation; for the pixels at the boundary of the image img2, the complete area covered by the convolution kernel is not existed, and the pixels at the boundary are directly considered as background pixels; assigning a corresponding pixel point on the gray image img1 as a gray background value avg1 (see B4); specifically:
performing corrosion operation on the pixel point p, wherein a convolution value y1 is a convolution value calculated at the pixel point p by using a convolution Kernel Kernel1, the value range of y1 is [0, 665] (when the value 29 is taken for edge 2), a corrosion threshold Nt1 is given, if Nt1=631 is set, if y1 is larger than Nt1, the p point on img2 is assigned as 1 (the p point is a spot, namely a spot foreground), otherwise, the p point is assigned as 0 (the p point is a background);
performing expansion operation on the pixel point p to obtain a convolution value y2, wherein the y2 calculation method is the same as that of y1 (see calculation of C2.2y1); given an expansion threshold Nt2, such as nt2=11, if y2 > Nt2, then the p point on img2 is assigned a value of 1 (blob), otherwise it is assigned a value of 0 (background);
C3. removing sand hole noise on the speckle foreground of the image img2 based on the mathematical morphology closed operation;
c3.1 sets the convolution Kernel2 as a square pixel matrix B with a size edge by edge (edge is odd, for example, edge=15), and the value of each pixel of the convolution Kernel2 is 1;
c3.2, performing expansion operation and then corrosion operation on each pixel point p on the binarized image img2 by using a convolution Kernel Kernel2 (the convolution Kernel origin is at the pixel point p), directly recognizing the pixel at the boundary of the image img2 as a background pixel if the pixel at the boundary is not a complete area covered by the convolution Kernel, and assigning a gray background value avg1 to the corresponding pixel point on the gray image img1 (see B4); specifically:
performing expansion operation on the pixel point p to obtain a convolution value y3, wherein the calculation method of the y3 is the same as that of y1, a threshold Nt3 is given, if Nt3=7, if y3 is larger than Nt3, a value of 1 (a spot) is assigned to a p point on img2, otherwise, a value of 0 (background) is assigned;
performing expansion operation on the pixel point p to obtain a convolution value y4, wherein the calculation method of the y4 is the same as that of y1, a threshold Nt4 is given, if Nt4=203 (when the value of the edge is 15), if y4 is larger than Nt4, a value of 1 (a spot) is assigned to a p point on img2, otherwise, a value of 0 (background) is assigned;
C4. removing background pores and filling foreground sand holes on the gray image img1; the method comprises the following operations:
c4.1 according to the binarization threshold Vt, the pore in the background is removed for the image img1 by: when the gray value of a point on the gray image img1 corresponding to the background pixel is smaller than a binary threshold Vt (pore, pseudo spot pixel, see C1), the pixel point in img1 is reassigned with the gray value of avg1 (background gray average, see B4) by taking the pixel with the pixel value of 0 on the binary image img2 as the background, so that the pore of the background is removed;
c4.2 filling sand holes in the speckle prospect for the image img1, wherein the method comprises the following steps: if the gray value of a point on the gray image img1 corresponding to the pixel is larger than a binary threshold Vt (sand hole, see C1), the gray value of the pixel point in img1 is reassigned to avg1-50 (seed point gray value in the spot, see E), so that the sand hole of the foreground spot is filled; avg1 is the average gray scale after stretching;
D. the continuous closed dividing line (contour line) of the spots is obtained on the gray image img1 by adopting a watershed algorithm, and the result is binary
Imaging img3;
setting a threshold mark, wherein mark=avg1-C2, C2 is a constant (for example, mark=avg1-50 is set), obtaining a continuous closed dividing line of a spot area on a gray image img1 through a watershed algorithm based on the threshold mark, wherein avg1 is a stretched gray average value (refer to B4), and obtaining a binary image img3 after processing through the watershed algorithm, wherein: 0 represents a continuously closed blob outline, 255 represents background;
E. for the binarized image img3, a seed filling algorithm of computer graphics is adopted to fill the inside of the spot dividing line, and the specific steps are as follows:
E1. determining a seed point, wherein a point with a pixel point gray value equal to a mark value on the gray image img1 is the seed point, and the mark is referred to as D;
E2. filling the inside of the spot;
on the binarized image img3, each continuous closed region is a spot block, at least one seed point (corresponding to a seed pixel point of the gray image img1, see E1) is arranged in each spot block, each spot block fills the inner region of the spot block by adopting an eight-communication seed filling algorithm, the pixel value of the inner region is 1, the 0 value of a spot dividing line is used for distinguishing, and the result image img4 obtained after the filling of img3 is a three-valued image, wherein: a value of 0 represents a continuous closed contour (split line) pixel of the blob, 1 represents an intra-blob pixel, and 255 represents a background pixel;
F. calculating the area, perimeter, roundness, center point and darkness of each spot by using the tri-valued image img4, wherein the method comprises the following steps:
f1, calculating center points of the outer rectangles (horizontal/vertical rectangle sides) A1 and A1 of each spot, wherein the center point of each outer rectangle is defined as the center point of each spot;
f2, defining the radius r of the spot circumscribing circle, wherein the maximum distance r of all pixel points in each spot from the center point of the spot is defined as the radius of the spot circumscribing circle;
f3 spot area num, the number num of all pixel points in each spot is defined as spot area;
f4 spot roundness=num/(pi×r×r);
f5 the perimeter cir of the spots, the number of pixels on each spot continuous closed contour is defined as the perimeter cir of the spot;
f6, defining the average value of gray values of pixels in each spot on the gray image img1 as spot darkness;
G. the spot size, spacing, number, darkness and duty ratio of the whole skin image are counted as follows
G1. Spot size, the average of all the Circularity values of the spots can characterize the spot size of the skin image;
G2. the distance between each spot and the nearest spot is d, and the average value of d values of all spots can represent the spot distance span, wherein the d value is obtained according to the central points of 2 spots;
G3. the number of spots on the whole image, the number of continuous closed contour lines is defined as the number of spots being count, and each continuous closed dividing line represents a spot;
G4. spot darkness, wherein the average value of darkness of all spots can represent the darkness dark of the spots of the whole picture;
G5. the speckle ratio, the ratio of the area of all the spots (the pixels of the spots) to the whole image (the pixels of all the spots), is defined as the speckle ratio.
The method of the invention realizes a set of skin detection system, and the skin detection evaluation system comprises skin imaging hardware equipment, a computer server side and a mobile phone client side; the skin image acquisition hardware equipment is used for acquiring micro-distance images, and acquiring images with the same resolution and consistent size; the mobile phone client is used for uploading the acquired image to the computer server and receiving and displaying the skin image attribute value returned by the computer server; the computer server side is used for quantitatively calculating a plurality of attribute values of the skin surface spots of each image.
In specific implementation, a micro-distance skin imaging device is adopted to acquire a skin image with 1000-1000 resolution; the computer server side installs Windows server 2012 and MySql5.7.16; and the mobile phone client adopts an Android smart mobile phone.
In specific implementation, the computer server side includes: the device comprises a skin gray image acquisition module, an image processing module, an image skin spot contour line acquisition module, an image skin spot contour line internal filling module and a skin spot attribute information calculation module; the skin gray image acquisition module is used for recombining three different color components in the RGB color space of the skin image to obtain a skin gray image; the image processing module is used for removing hairs and white noise in the skin gray image, unifying the illumination intensity of the image, and further performing treatment of removing pore and sand hole noise; the image skin spot contour line acquisition module is used for processing the skin gray image by adopting a computer image watershed algorithm to obtain a continuous closed parting line (contour line) of skin spots in the image; the image skin spot contour line internal filling module fills the interior of a continuous closed dividing line of the skin spot in the image through a computer graphics seed filling algorithm; the skin spot attribute information calculation module is used for calculating the area, perimeter, roundness, central point and darkness of each spot of the skin in the image, and the size, quantity, darkness, interval, duty ratio and other information of each skin spot.
Compared with the prior art, the invention has the beneficial effects that:
the invention provides a quantitative evaluation method of skin image spots based on a watershed and a seed filling algorithm, which is used for calculating a plurality of attribute characteristic values of surface spots of an image according to pixel color value information of the skin image so as to realize detection and quantitative evaluation of the skin image spots. The method has the advantages that the RGB color space pixel color values are used for detecting the spot characteristic values, the skin image is the only parameter, the mathematical morphology, the computer image algorithm and the computer graphics algorithm are used for calculating the skin spot indexes, and the method is high in accuracy, high in speed and good in market application prospect and value in detecting spots and quantitatively evaluating the spots.
Drawings
FIG. 1 is a block diagram of a quantitative evaluation system for skin image speckle embodying the present invention.
FIG. 2 is a block diagram of a quantitative evaluation flow of skin image spots provided by the invention.
FIG. 3 is a flow chart of the method for quantitatively calculating skin image spots provided by the invention.
FIG. 4 is a schematic representation of the gray value range of a patch pixel of a skin image when the present invention is implemented.
FIG. 5 is a schematic representation of a convolution template employed in processing a skin image in the practice of the present invention.
FIG. 6 is an example of a partial skin image for calculating skin spots in accordance with an embodiment of the present invention.
Detailed Description
The invention is further illustrated by the following examples in conjunction with the accompanying drawings, but in no way limit the scope of the invention.
The invention provides a quantitative evaluation method of skin image spots based on a watershed and a seed filling algorithm, which is characterized in that according to pixel color value information of RGB color space of a skin image, skin micro-distance digital images with the same resolution are processed, a plurality of attribute characteristic values of the surface spots of the skin image are calculated, the attribute characteristic values are a plurality of attribute index quantitative numerical values for measuring a skin image spot, and the numerical values can identify the spot characteristics of the skin surface of the micro-distance image, so that quantitative evaluation of the skin image spots is realized. The method mainly comprises the following steps:
1) Recombining three different color components in the RGB color space of the skin image to obtain a skin gray image;
2) Removing hairs and white noise in the image from the skin gray image, and unifying the illumination intensity of the image;
3) Further processing the skin gray image obtained in the step 2) to remove pores and sand hole noise;
4) Processing the skin gray image obtained in the step 3) by using a computer image watershed algorithm to obtain a continuous closed parting line (contour line) of skin spots in the image;
5) Filling the interior of a continuous closed dividing line of skin spots in the image by using a computer graphics seed filling algorithm;
6) Calculating the area, perimeter, roundness, center point and darkness of a single spot of the skin in the image;
7) And counting to obtain the size, the interval, the number, the darkness and the duty ratio of the skin spots in the whole image.
Specifically, the method of the present invention comprises the steps of:
A. the recombination of three different color components of the skin image yields a gray image img1, which is specifically as follows:
A1. reading a skin image file into a memory;
the only parameter of the algorithm is an RGB space image file stored on a hard disk, and the file is read into a memory;
A2. graying the color image to obtain a gray image img1;
the three different color components of the RGB image are weighted and averaged to obtain a gray image img1, and the weighted average calculation method is as follows: g=r×0.7+g×0.1+b×0.2, where G is a gray value of a pixel, R, G, B is red, green, and blue color components of the pixel, and a gray value G value range is [0,255];
B. on the gray image img1, hair and white noise are removed, and the illumination intensity is unified, and the specific steps are as follows:
B1. calculating a gray image img1 gray average avg, wherein avg=the sum of all pixel gray values and dividing the sum by the number of pixels;
B2. removing hair;
the hair pixels are represented on the gray image img1 as small gray values, the color is almost black, a hair threshold v1 (v1=40+avg-127) is given, and all pixel points smaller than the threshold are reassigned to the background color avg on the image img1;
B3. white noise is removed;
white noise appears as a large gray value in the gray image img1, the color is almost white, a white noise threshold v2 (v2=220+avg-127) is given, and all pixel points with gray values larger than the threshold are reassigned to be background color avg on the image img1;
B4. unifying the illumination intensity;
after hair and white noise are removed, the pixel gray value interval of the gray image img1 is [ v1, v2], the interval is stretched to interval [0,255] in proportion, and the uniform illumination intensity is used for avoiding calculation errors (for comparing different pictures) caused by illumination differences of a plurality of skin images, and the calculation method is as follows:
G1＝(G－v1)×255/(v2－v1)，
wherein G is a pixel gray value of img1, v1 and v2 are respectively hair and white noise threshold values, the gray average avg is avg1 (stretched gray average), avg1= (avg-v 1) ×255/(v 2-v 1);
C. pore and sand hole noise are removed from the gray image img1, and the specific contents are as follows:
C1. binarizing the gray level image to obtain a binarized image img2;
determining a binarization threshold Vt, wherein vt=avg1-30, and obtaining a value b of each pixel point on the binarized image img2, wherein the value b of each pixel point on the binarized image img2 is determined according to a gray value v of a corresponding point of the pixel point on the gray image img1, and the method specifically comprises the following steps:
c1.1 if v > Vt, b=0, background+speckle foreground sand hole noise point;
c1.2 if v < = Vt, b = 1, pores in the foreground + background of the sand holes.
C2. Removing pores in an image img2 background based on a mathematical morphology open operation method;
c2.1, setting a convolution Kernel Kernel1 based on a mathematical morphology open operation method; kernel1 is a square matrix a of pixels of size edge2 x edge2 (edge 2 is an odd number, e.g., edge 2=29); in the square pixel matrix A, the center point of the square is a convolution kernel origin, the convolution kernel pixel value is 1 or 0, the pixel value of an inscribed circle part of the square is 1, the pixel value outside the inscribed circle is 0; the convolution kernel (square pixel matrix a) has a total of 665 1 values, 176 0 values;
c2.2, regarding each pixel point p on the binarized image img2, setting the position of the pixel point p as a convolution Kernel origin, performing image corrosion operation by using a convolution Kernel Kernel1, and performing image expansion operation; for the pixels at the boundary of the image img2, the complete area covered by the convolution kernel is not existed, and the pixels at the boundary are directly considered as background pixels; assigning a corresponding pixel point on the gray image img1 as a gray background value avg1 (see B4); specifically:
performing corrosion operation on the pixel point p, wherein a convolution value y1 is a convolution value calculated at the pixel point p by using a convolution Kernel Kernel1, the value range of y1 is [0, 665], a corrosion threshold Nt1 is set if Nt1=631, if y1 is larger than Nt1, the p point on img2 is assigned as 1 (the p point is a spot, namely a spot foreground), otherwise, the p point is assigned as 0 (the p point is a background);
performing expansion operation on the pixel point p to obtain a convolution value y2, wherein the y2 calculation method is the same as that of y1 (see calculation of C2.2y1); given an expansion threshold Nt2, such as nt2=11, if y2 > Nt2, then the p point on img2 is assigned a value of 1 (blob), otherwise it is assigned a value of 0 (background);
C3. removing sand hole noise on the speckle foreground of the image img2 based on the mathematical morphology closed operation;
c3.1 sets the convolution Kernel2 as a square pixel matrix B with a size edge by edge (edge is odd, for example, edge=15), and the value of each pixel of the convolution Kernel2 is 1;
c3.2, performing expansion operation and then corrosion operation on each pixel point p on the binarized image img2 by using a convolution Kernel Kernel2 (the convolution Kernel origin is at the pixel point p), directly recognizing the pixel at the boundary of the image img2 as a background pixel if the pixel at the boundary is not a complete area covered by the convolution Kernel, and assigning a gray background value avg1 to the corresponding pixel point on the gray image img1 (see B4); specifically:
performing expansion operation on the pixel point p to obtain a convolution value y3, wherein the calculation method of the y3 is the same as that of y1, a threshold Nt3 is given, if Nt3=7, if y3 is larger than Nt3, a value of 1 (a spot) is assigned to a p point on img2, otherwise, a value of 0 (background) is assigned;
performing expansion operation on the pixel point p to obtain a convolution value y4, wherein the calculation method of the y4 is the same as that of y1, a threshold Nt4 is given, if Nt4=203, if y4 is larger than Nt4, a value of 1 (a spot) is assigned to a point p on img2, otherwise, a value of 0 (background) is assigned;
C4. removing background pores and filling foreground sand holes on the gray image img1; the method comprises the following operations:
c4.1 according to the binarization threshold Vt, the pore in the background is removed for the image img1 by: when the gray value of a point on the gray image img1 corresponding to the background pixel is smaller than a binary threshold Vt (pore, pseudo spot pixel, see C1), the pixel point in img1 is reassigned with the gray value of avg1 (background gray average, see B4) by taking the pixel with the pixel value of 0 on the binary image img2 as the background, so that the pore of the background is removed;
c4.2 filling sand holes in the speckle prospect for the image img1, wherein the method comprises the following steps: if the gray value of a point on the gray image img1 corresponding to the pixel is larger than a binary threshold Vt (sand hole, see C1), the gray value of the pixel point in img1 is reassigned to avg1-50 (seed point gray value in the spot, see E), so that the sand hole of the foreground spot is filled; avg1 is the average gray scale after stretching;
D. a watershed algorithm is adopted on the gray level image img1 to obtain a continuous closed dividing line (contour line) of the spots, and the result is a binarized image img3;
giving a threshold mark on a gray image img1, realizing a watershed algorithm based on the threshold mark on img1 to obtain a spot area continuous closed dividing line, setting mark=avg1-50, wherein avg1 is a stretched gray average value (refer to B4), and obtaining a binarized image img3 after processing by the watershed algorithm, wherein: 0 represents a continuously closed blob outline, 255 represents background;
E. for the binarized image img3, a seed filling algorithm of computer graphics is adopted to fill the inside of the spot dividing line, and the specific steps are as follows:
E1. determining a seed point, wherein a point with a pixel point gray value equal to a mark value on the gray image img1 is the seed point, and the mark is referred to as D;
E2. filling the inside of the spot;
on the binarized image img3, each continuous closed region is a spot block, at least one seed point (corresponding to the seed pixel point of the gray image img1, see E1) is arranged in each spot block, each spot block is respectively filled with an inner region of the spot block by using an eight-connected seed filling algorithm, the pixel value of the inner region is 1 to distinguish the 0 value of a spot dividing line, and the result image img4 after the filling of img3 is a three-valued image, wherein: a value of 0 represents a continuous closed contour (split line) pixel of the blob, 1 represents an intra-blob pixel, and 255 represents a background pixel;
F. calculating the area, perimeter, roundness, center point and darkness of each spot by using the tri-valued image img4, wherein the method comprises the following steps:
f1, calculating center points of the outer rectangles (horizontal/vertical rectangle sides) A1 and A1 of each spot, wherein the center point of each outer rectangle is defined as the center point of each spot;
f2, defining the radius r of the spot circumscribing circle, wherein the maximum distance r of all pixel points in each spot from the center point of the spot is defined as the radius of the spot circumscribing circle;
f3 spot area num, the number num of all pixel points in each spot is defined as spot area;
f4 spot roundness=num/(pi×r×r);
f5 the perimeter cir of the spots, the number of pixels on each spot continuous closed contour is defined as the perimeter cir of the spot;
f6, defining the average value of gray values of pixels in each spot on the gray image img1 as spot darkness;
G. the spot size, spacing, number, darkness and duty ratio of the whole skin image are counted as follows
G1. Spot size, the average of all the Circularity values of the spots can characterize the spot size of the skin image;
G2. the distance between each spot and the nearest spot is d, and the average value of d values of all spots can represent the spot distance span, wherein the d value is obtained according to the central points of 2 spots;
G3. the number of spots on the whole image, the number of continuous closed contour lines is defined as the number of spots being count, and each continuous closed dividing line represents a spot;
G4. spot darkness, wherein the average value of darkness of all spots can represent the darkness dark of the spots of the whole picture;
G5. the speckle ratio, the ratio of the area of all the spots (the pixels of the spots) to the whole image (the pixels of all the spots), is defined as the speckle ratio.
The method of the invention realizes a set of skin detection system, and the skin detection evaluation system comprises skin imaging hardware equipment, a computer server side and a mobile phone client side; the skin image acquisition hardware equipment is used for acquiring micro-distance images, and acquiring images with the same resolution and consistent size; the mobile phone client is used for uploading the acquired image to the computer server and receiving and displaying the skin image attribute value returned by the computer server; the computer server side is used for quantitatively calculating a plurality of attribute values of the skin surface spots of each image.
In practice, the system configuration is as in table 1: acquiring a skin image with 1000 x 1000 resolution by using a micro-distance skin imaging device; the computer server side installs Windows server 2012 and MySql5.7.16; and the mobile phone client adopts an Android smart mobile phone.
TABLE 1 Equipment configuration of skin image detection evaluation System according to an embodiment of the present invention
| Name of the name | Device model | Quantity of |
| Skin image capturing device | Micro-distance skin imaging equipment for acquiring skin images with resolution of 1000 x 1000 | 6 |
| Cloud server | Windows server 2012、MySql5.7.16 | 1 |
| Client terminal | Android client of mobile phone | 6 |
The skin detection and evaluation system consists of skin imaging hardware equipment, a computer server side and a mobile phone client side, wherein the structural block diagram is shown in fig. 1, the skin quantitative evaluation flow is shown in fig. 2, and the skin detection and evaluation system specifically comprises the following implementation steps: (1) The special skin image capturing equipment is used for capturing images with the same resolution, the sizes of the images are consistent, and the consistent pretreatment of the sizes of the skin images is avoided; (2) The acquired image is uploaded to a server through a mobile phone client; (3) The server side quantitatively calculates a plurality of attribute values of the skin surface spots of each image; (4) The skin image attribute values are returned to the mobile client and displayed.
The quantitative evaluation method of skin spots provided by the invention only needs one image as a calculation data source, and the specific implementation steps and the invention content are as follows, and the embodiment flow steps are as follows, referring to fig. 3:
1) The server reads the image uploaded by the mobile phone client into the memory;
2) Graying the image to obtain a gray image img1;
3) Preprocessing a gray image img 1: according to the gray value on the image img1, the pixels which are too bright (white noise) and too dark (hair) are removed by a simple threshold method, the gray range of the image is stretched in proportion, and the illumination intensities of different images are unified;
4) The gray image img1 is binarized, in this embodiment, the binarization threshold is the average value of the gray image img1 minus 30 (see fig. 4), so as to obtain a binarized image img2, wherein the 0 value is mainly the background pixel (background+front Jing Shayan), the 1 value is mainly the foreground pixel (spot+pore), and the gray image binarization schematic diagram is shown in fig. 4;
5) The background pores are removed by a mathematical morphology-based open operation method, wherein the convolution kernel is shown in fig. 5, the convolution value threshold in the open operation is adjusted, the corrosion operation threshold is 631, and the expansion operation threshold is 11;
6) Filling foreground sand hole noise based on a mathematical morphology closed operation method, and adjusting a convolution value threshold in the closed operation, wherein the expansion threshold is 203, and the corrosion threshold is 7;
7) A continuous closed dividing line (contour line) of the spot is obtained based on a mark (taking a mark value as a gray image mean value minus 50, see fig. 4) computer image watershed algorithm, and the result is a binarized image img3, wherein a value 0 is the spot dividing line, and a value 255 is the background;
8) The computer graphics seed filling algorithm fills the inside of the spot dividing line, the result is a three-valued image img3, wherein 0 value is a spot contour line, 255 is background, 1 is a pixel inside the spot, and a seed point schematic diagram related to the seed filling algorithm is shown in fig. 4;
9) Calculating the area, perimeter, roundness, central point and darkness of a single spot;
10 Size, spacing, number, darkness, and duty cycle of the entire image spot.
The embodiment results show that the method for detecting and evaluating the skin surface spots based on the image RGB space is high in calculation speed and high in accuracy of detection results. In this embodiment, spot evaluation calculation is performed on 89 images, specifically, 6 different image capturing devices capture micro-distance skin images and upload the images to a server, the server calculates the spot attribute value of each image by adopting the method of the present invention, fig. 6 is a part of the images, the corresponding spot attribute calculation result values are shown in table 2, and all other values except the number index are percentage values, that is, the value range is [0,100].
TABLE 2 Spot attribute values for each image calculated by the method of the present invention
It should be noted that the purpose of the disclosed embodiments is to aid further understanding of the present invention, but those skilled in the art will appreciate that: various alternatives and modifications are possible without departing from the spirit and scope of the invention and the appended claims. Therefore, the invention should not be limited to the disclosed embodiments, but rather the scope of the invention is defined by the appended claims.
Claims (10)
1. A skin image speckle assessment method based on watershed and seed filling uses an acquired skin image as a unique data source, calculates quantitative values of a plurality of attributes of skin speckle, and can identify speckle characteristics of the skin surface of the image, thereby realizing skin image speckle assessment; the method comprises the following steps:
1) Graying the skin image by utilizing different color components in the skin image color space to obtain a skin gray image;
specifically, carrying out weighted average on RGB color components of each pixel of the skin image in a color space to obtain a gray value of each pixel, and obtaining a corresponding gray image img1;
2) Removing hairs and white noise in the image from the skin gray image, and unifying the illumination intensity of the image; the method specifically comprises the following steps:
B1. calculating the average avg of the gray values of the gray image img1 as a background color;
B2. removing hair: setting a hair threshold v1, and reassigning all pixel points with gray values smaller than the hair threshold on the gray image img1 to be values avg of background colors;
B3. white noise is removed: setting a white noise threshold v2, v1< v2; reassigning all pixel points with gray values larger than a white noise threshold value on the gray image processed by the B2 to be values avg of background colors; the pixel gray value interval of the skin gray image after hair and white noise are removed is [ v1, v2];
B4. unifying the illumination intensity: stretching the pixels of the skin gray image with the hair removed and the white noise removed to an interval [0,255] in proportion; the gray average value is stretched to be the stretched gray average value avg1;
3) Further processing the skin gray image processed in the step 2) to remove pores and sand hole noise; the method specifically comprises the following steps:
C1. binarizing the skin gray level image processed in the step 2) to obtain a binarized image img2;
determining a binarization threshold Vt, wherein Vt is smaller than avg1-C, and C is a constant to obtain a binarization image img2; the value b of each pixel point on img2 is determined according to the gray value v of the corresponding point of the pixel point on the skin gray image processed in the step 2), if v is larger than Vt, b=0, and the pixel point is identified as a sand hole noise point in the background or the speckle foreground; if v < = Vt, b = 1, identifying the pixel as a pore in the foreground or background of the blob;
C2. removing pores in a binary image img2 background obtained after C1 processing based on an open operation method; the method specifically comprises the following steps:
c2.1 sets a convolution Kernel1 for image processing; the convolution kernel is a weight definition function adopted in a pixel of an output image obtained by carrying out weighted average on the pixel in a small area in the input image; setting Kernel1 as a square pixel matrix A, wherein the center point of the square is a convolution Kernel origin, the convolution Kernel pixel value is 1 or 0, the pixel value of an inscribed circle part of the square is 1, and the pixel value outside the inscribed circle and in the square is 0;
c2.2, setting the p point position as a convolution Kernel origin point for each pixel point p on the binarized image obtained after the C1 processing, and performing image corrosion operation and then image expansion operation by using a convolution Kernel1; the pixels at the boundary of the binarized image obtained after C1 processing are not a complete area covered by a convolution kernel, and are directly identified as background pixels; assigning a gray background value avg1 to the corresponding pixel point on the skin gray image processed in the step 2);
C3. removing sand hole noise on the speckle foreground of the binarized image processed by the C2 based on a closed operation method; the method specifically comprises the following steps:
c3.1, setting a convolution Kernel Kernel2 as a square pixel matrix B, wherein the value of each pixel of the convolution Kernel Kernel2 is 1;
c3.2, setting the origin of the convolution Kernel Kernel2 at each pixel point p on the binarized image processed by the C2, and performing expansion operation on the pixel point p by using the convolution Kernel Kernel2 and then performing corrosion operation; recognizing pixels in a complete area, which is not covered by a convolution kernel, of pixels at the boundary of the image img2 as background pixels, and assigning a gray background value avg1 to corresponding pixel points on the skin gray image processed in the step 2);
C4. removing background pores and filling foreground sand holes on the skin gray image treated by the C3; the method comprises the following operations:
c4.1, according to the binarization threshold Vt, removing pores in the background of the gray level image processed by the step 2), wherein the method comprises the following steps: taking a pixel with a pixel value of 0 on the binarized image img2 as a background, and when the gray value of a point on the gray image corresponding to the background pixel is smaller than a binarization threshold Vt, re-assigning the gray value of the pixel point in the gray image with an avg1, thereby removing pores of the background;
c4.2 filling sand holes in the speckle prospect for the gray level image processed by the C4.1, wherein the method comprises the following steps: if the gray value of a point on the gray image corresponding to the pixel is larger than a binary threshold Vt, reassigning the gray value of the pixel point in the gray image to avg1-50, thereby filling the sand holes of the foreground spots; avg1 is the average gray scale after stretching;
4) Processing the skin gray image obtained in the step 3) by utilizing an image watershed algorithm to obtain a continuous closed dividing line, namely a contour line, of skin spots in the image;
specifically, setting a threshold mark, wherein mark=avg1-C2, C2 is a constant, and obtaining a continuous closed dividing line of a spot area on the gray level image processed in the step 3) through a watershed algorithm based on the threshold mark, thereby obtaining a binarized image img3, wherein: 0 represents a continuously closed blob outline, 255 represents background;
5) Filling the interior of the continuous closed dividing line of the skin spots in the image by using a seed filling algorithm; the method specifically comprises the following steps:
E1. determining a seed point: taking the point with the pixel point gray value equal to the threshold mark on the gray image processed in the step 3) as a seed point;
E2. filling the inside of the spot: on the binarized image img3 processed in the step 4), each continuous closed area is a spot block, and at least one seed point is arranged in each spot block; filling the inner area of each spot block respectively, wherein the pixel value of the inner area is 1, which is different from the pixel value 0 of the spot dividing line;
obtaining a result image img4 after filling, wherein img4 is a three-valued image, and the three-valued image comprises the following components: a value of 0 represents a continuous closed contour pixel of the blob, 1 represents a pixel inside the blob, and 255 represents a background pixel;
6) Calculating to obtain the area, perimeter, roundness, center point and darkness of a single spot of the skin in the ternary image obtained by the treatment in the step 5);
7) Counting and calculating to obtain the size, the interval, the number, the darkness and the duty ratio of the skin spots in the ternary image obtained by the treatment in the step 5);
through the above steps, evaluation of skin image spots based on watershed and seed filling is achieved.
2. The method for evaluating skin image spots based on watershed and seed filling as in claim 1, wherein the computing method of step 6) specifically comprises:
F1. determining the center point of the outsourcing rectangle A1 of each spot as the center point of each spot;
F2. calculating to obtain the maximum distance between all pixel points in each spot and the center point of the spot, and taking the maximum distance as the radius r of the spot circumscribing circle;
F3. taking the number num of all pixel points in each spot as the spot area;
F4. the spot roundness Circularity is calculated by the formula circularity=num/(pi×r×r);
F5. calculating to obtain the number of pixels on the continuous closed contour line of each spot, and taking the number as the spot circumference cir;
F6. and (3) calculating a gray average value of pixels in each spot corresponding to the gray image processed in the step (3) to be used as the darkness of the spot.
3. The method for evaluating skin image spots based on watershed and seed filling according to claim 2, wherein the step 7) specifically comprises:
G1. taking the average value of roundness values of all spots as the size of the skin image spots;
G2. the distance between each spot and the nearest spot is d, and the d value is obtained according to the center points of 2 spots; taking the average value of d values of all the spots as a spot distance span;
G3. each successive closed dividing line in the image represents a blob; the number of the continuous closed contour lines is the spot number count;
G4. taking the darkness average value of all the spots as a spot darkness dark of the image;
G5. the ratio of the area of all spots to all pixels in the image is taken as the spot ratio.
4. The watershed and seed filling-based skin image spot assessment method according to claim 1, wherein the skin image is a macro image.
5. The method for evaluating skin image spots based on watershed and seed filling as claimed in claim 1, wherein in step B2, a white noise threshold v2=220+avg-127 is set.
6. The watershed and seed filling-based skin image spot assessment method according to claim 1, wherein the square pixel matrix a of the convolution Kernel1 has 665 1 values and 176 0 values in total; the square pixel matrix B of the convolution Kernel2 has a square pixel matrix B of 15 x 15, and each pixel has a value of 1.
7. The method for evaluating skin image spots based on watershed and seed filling according to claim 1, wherein in step C2.2, specifically: performing corrosion operation on the pixel point, wherein a convolution value calculated at the pixel point p by using a convolution Kernel Kernel1 is y1, and the value range of y1 is [0, 665]; setting an erosion threshold Nt1, and if y1 is larger than Nt1, assigning 1 to the p point on the binary image obtained after C1 treatment, wherein the p point is a spot, namely a spot prospect; otherwise, the p point is assigned to 0, which means that the p point is the background;
performing expansion operation on the pixel point p to obtain a convolution value y2; setting an expansion threshold Nt2, and if y2 is larger than Nt2, assigning a value of 1 to a point p on the binarized image, wherein the point p is a spot; otherwise, a value of 0 is assigned, indicating that the point is background.
8. The method for evaluating a speckle of a skin image based on watershed and seed filling as in claim 7, wherein step E2 fills the interior region of the speckle block by an eight-way seed filling algorithm.
9. A skin image spot evaluation detection system based on the skin image spot evaluation method based on watershed and seed filling according to any one of claims 1-8, comprising a skin imaging hardware device, a computer server side and a mobile phone client side; the skin image acquisition hardware equipment is used for acquiring micro-distance images, and acquiring images with the same resolution; the mobile phone client is used for uploading the acquired images to the computer server and receiving and displaying skin image attribute values returned by the computer server; the computer server side is used for quantitatively calculating a plurality of attribute values of the skin surface spots of each image;
the computer server side comprises: the device comprises a skin gray image acquisition module, an image processing module, an image skin spot contour line acquisition module, an image skin spot contour line internal filling module and a skin spot attribute information calculation module; the skin gray image acquisition module is used for processing three color components in the RGB color space of the skin image to obtain a skin gray image; the image processing module is used for removing hairs and white noise in the image for the skin gray level image, unifying the illumination intensity of the image and removing pore and sand hole noise; the image skin spot contour line acquisition module processes the skin gray image by adopting a watershed algorithm to obtain a continuous closed parting line, namely a contour line, of the skin spot in the image; the image skin spot contour line internal filling module fills the inside of a continuous closed dividing line of the skin spot in the image through a seed filling algorithm; the skin spot attribute information calculating module is used for calculating various attribute information of each skin spot in the image, and comprises the following steps: area, circumference, roundness, center point, darkness, size, number, darkness and spacing, duty cycle information of each skin spot.
10. The skin image speckle evaluation detection system of claim 9, wherein the skin image acquired with the macro skin imaging device is a 1000 x 1000 resolution skin image; and/or, the computer server installs Windows server 2012 and MySql5.7.16; and/or the mobile phone client adopts an Android smart mobile phone.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010473389.7A CN111738984B (en) | 2020-05-29 | 2020-05-29 | Skin image spot evaluation method and system based on watershed and seed filling |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010473389.7A CN111738984B (en) | 2020-05-29 | 2020-05-29 | Skin image spot evaluation method and system based on watershed and seed filling |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111738984A CN111738984A (en) | 2020-10-02 |
| CN111738984B true CN111738984B (en) | 2023-08-18 |
Family
ID=72647953
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010473389.7A Active CN111738984B (en) | 2020-05-29 | 2020-05-29 | Skin image spot evaluation method and system based on watershed and seed filling |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111738984B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113379680B (en) * | 2021-05-18 | 2023-05-23 | 上海闻泰信息技术有限公司 | Defect detection method, defect detection device, electronic device and computer readable storage medium |
| CN113554668B (en) * | 2021-07-27 | 2022-02-22 | 深圳大学 | Skin mirror image melanoma segmentation method, device and related components |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105787929A (en) * | 2016-02-15 | 2016-07-20 | 天津大学 | Skin rash point extraction method based on spot detection |
| CN105869155A (en) * | 2016-03-24 | 2016-08-17 | 天津大学 | Face skin reddening region extraction method |
| CN106384075A (en) * | 2016-03-11 | 2017-02-08 | 株式会社爱茉莉太平洋 | A skin block mass based skin texture evaluating device and evaluating method |
| CN107072597A (en) * | 2014-11-19 | 2017-08-18 | 株式会社资生堂 | Spot apparatus for evaluating, spot appraisal procedure and program |
| CN107169960A (en) * | 2017-05-15 | 2017-09-15 | 精诚工坊电子集成技术（北京）有限公司 | A kind of skin surface pore size detection method based on color of image space |
| CN107157447A (en) * | 2017-05-15 | 2017-09-15 | 精诚工坊电子集成技术（北京）有限公司 | The detection method of skin surface roughness based on image RGB color |
| CN109859257A (en) * | 2019-02-25 | 2019-06-07 | 北京工商大学 | A kind of skin image texture appraisal procedure and system based on grain direction |
| CN110197487A (en) * | 2019-03-06 | 2019-09-03 | 上海商路网络科技有限公司 | A method of color spot is detected based on skin image |
| WO2020015148A1 (en) * | 2018-07-16 | 2020-01-23 | 华为技术有限公司 | Skin spot detection method and electronic device |
| CN111047619A (en) * | 2018-10-11 | 2020-04-21 | 展讯通信（上海）有限公司 | Face image processing method and device and readable storage medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| FR2891641B1 (en) * | 2005-10-04 | 2007-12-21 | Lvmh Rech | METHOD AND APPARATUS FOR CHARACTERIZING SKIN IMPERFECTIONS AND METHOD OF ASSESSING THE ANTI-AGING EFFECT OF A COSMETIC PRODUCT |
-
2020
- 2020-05-29 CN CN202010473389.7A patent/CN111738984B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107072597A (en) * | 2014-11-19 | 2017-08-18 | 株式会社资生堂 | Spot apparatus for evaluating, spot appraisal procedure and program |
| CN105787929A (en) * | 2016-02-15 | 2016-07-20 | 天津大学 | Skin rash point extraction method based on spot detection |
| CN106384075A (en) * | 2016-03-11 | 2017-02-08 | 株式会社爱茉莉太平洋 | A skin block mass based skin texture evaluating device and evaluating method |
| CN105869155A (en) * | 2016-03-24 | 2016-08-17 | 天津大学 | Face skin reddening region extraction method |
| CN107169960A (en) * | 2017-05-15 | 2017-09-15 | 精诚工坊电子集成技术（北京）有限公司 | A kind of skin surface pore size detection method based on color of image space |
| CN107157447A (en) * | 2017-05-15 | 2017-09-15 | 精诚工坊电子集成技术（北京）有限公司 | The detection method of skin surface roughness based on image RGB color |
| WO2020015148A1 (en) * | 2018-07-16 | 2020-01-23 | 华为技术有限公司 | Skin spot detection method and electronic device |
| CN111047619A (en) * | 2018-10-11 | 2020-04-21 | 展讯通信（上海）有限公司 | Face image processing method and device and readable storage medium |
| CN109859257A (en) * | 2019-02-25 | 2019-06-07 | 北京工商大学 | A kind of skin image texture appraisal procedure and system based on grain direction |
| CN110197487A (en) * | 2019-03-06 | 2019-09-03 | 上海商路网络科技有限公司 | A method of color spot is detected based on skin image |
Non-Patent Citations (1)
| Title |
|---|
| 一种基于人脸皮肤图像的色斑检测算法;储霞 等;《图像处理》;第25卷(第7-3期);第249-251页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111738984A (en) | 2020-10-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108323204B (en) | Method for detecting face flaw point and intelligent terminal | |
| CN108549874B (en) | Target detection method, target detection equipment and computer-readable storage medium | |
| CN107316077B (en) | Automatic adipose cell counting method based on image segmentation and edge detection | |
| CN110322522B (en) | Vehicle color recognition method based on target recognition area interception | |
| CN107169960B (en) | Skin surface pore size detection method based on image color space | |
| CN111915704A (en) | Apple hierarchical identification method based on deep learning | |
| CN115018844B (en) | Plastic film quality evaluation method based on artificial intelligence | |
| CN108898132B (en) | Terahertz image dangerous article identification method based on shape context description | |
| CN109087286A (en) | A kind of detection method and application based on Computer Image Processing and pattern-recognition | |
| CN109859257B (en) | Skin image texture evaluation method and system based on texture directionality | |
| CN111161222B (en) | Printing roller defect detection method based on visual saliency | |
| CN115082451B (en) | Stainless steel soup ladle defect detection method based on image processing | |
| CN111738984B (en) | Skin image spot evaluation method and system based on watershed and seed filling | |
| CN106442556A (en) | Device and method for detecting surface defects of perforated plate workpiece | |
| CN115205194B (en) | Image processing-based method, system and device for detecting coverage rate of armyworm plate | |
| CN111784675A (en) | Method and device for processing article texture information, storage medium and electronic equipment | |
| CN115908774B (en) | Quality detection method and device for deformed materials based on machine vision | |
| CN111709305B (en) | Face age identification method based on local image block | |
| CN107239761A (en) | Fruit tree branch pulling effect evaluation method based on skeleton Corner Detection | |
| KR100488014B1 (en) | YCrCb color based human face location detection method | |
| CN109949273B (en) | Skin image texture segmentation method and system based on texture symmetry | |
| CN111402189B (en) | Video image color cast detection device and method | |
| Long et al. | An Efficient Method For Dark License Plate Detection | |
| CN109087289A (en) | A kind of plate visual pattern detection algorithm under grid background | |
| US20200213582A1 (en) | Evaluation of dynamic ranges of imaging devices |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |