CN117274673A - Correlation calculation method of human face and macaque face based on acne - Google Patents
Correlation calculation method of human face and macaque face based on acne Download PDFInfo
- Publication number
- CN117274673A CN117274673A CN202311127893.1A CN202311127893A CN117274673A CN 117274673 A CN117274673 A CN 117274673A CN 202311127893 A CN202311127893 A CN 202311127893A CN 117274673 A CN117274673 A CN 117274673A
- Authority
- CN
- China
- Prior art keywords
- face
- image
- acne
- calculating
- area
- 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.)
- Pending
Links
- 208000002874 Acne Vulgaris Diseases 0.000 title claims abstract description 54
- 206010000496 acne Diseases 0.000 title claims abstract description 54
- 241000282553 Macaca Species 0.000 title claims abstract description 27
- 241000282414 Homo sapiens Species 0.000 title claims abstract description 17
- 238000004364 calculation method Methods 0.000 title claims abstract description 17
- 230000011218 segmentation Effects 0.000 claims abstract description 11
- 230000004927 fusion Effects 0.000 claims abstract description 9
- 238000012545 processing Methods 0.000 claims abstract description 6
- 238000013528 artificial neural network Methods 0.000 claims abstract description 4
- 241000510093 Quadrula metanevra Species 0.000 claims description 29
- 230000001815 facial effect Effects 0.000 claims description 18
- 230000010287 polarization Effects 0.000 claims description 15
- 238000001228 spectrum Methods 0.000 claims description 15
- 238000012549 training Methods 0.000 claims description 14
- 238000000034 method Methods 0.000 claims description 12
- 238000002329 infrared spectrum Methods 0.000 claims description 11
- 238000012360 testing method Methods 0.000 claims description 10
- 238000013527 convolutional neural network Methods 0.000 claims description 8
- 238000011282 treatment Methods 0.000 claims description 8
- 239000012855 volatile organic compound Substances 0.000 claims description 8
- 238000002372 labelling Methods 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 6
- 238000003062 neural network model Methods 0.000 claims description 5
- 210000000056 organ Anatomy 0.000 claims description 4
- 238000012935 Averaging Methods 0.000 claims description 3
- 206010061218 Inflammation Diseases 0.000 claims description 3
- 238000001514 detection method Methods 0.000 claims description 3
- 238000010586 diagram Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 210000001508 eye Anatomy 0.000 claims description 3
- 230000004054 inflammatory process Effects 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 210000000697 sensory organ Anatomy 0.000 claims description 3
- 230000006870 function Effects 0.000 claims description 2
- 230000037311 normal skin Effects 0.000 claims description 2
- 238000007781 pre-processing Methods 0.000 claims description 2
- 208000017520 skin disease Diseases 0.000 claims description 2
- 210000004209 hair Anatomy 0.000 abstract description 7
- 238000013135 deep learning Methods 0.000 abstract description 5
- 238000004519 manufacturing process Methods 0.000 abstract description 3
- 238000013461 design Methods 0.000 abstract description 2
- 238000011156 evaluation Methods 0.000 description 4
- 210000000887 face Anatomy 0.000 description 4
- 230000008506 pathogenesis Effects 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 208000032544 Cicatrix Diseases 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 2
- 230000002068 genetic effect Effects 0.000 description 2
- 231100000241 scar Toxicity 0.000 description 2
- 230000037387 scars Effects 0.000 description 2
- 238000013519 translation Methods 0.000 description 2
- 208000020154 Acnes Diseases 0.000 description 1
- 208000019901 Anxiety disease Diseases 0.000 description 1
- 241000186427 Cutibacterium acnes Species 0.000 description 1
- 201000004624 Dermatitis Diseases 0.000 description 1
- 206010047642 Vitiligo Diseases 0.000 description 1
- 230000036506 anxiety Effects 0.000 description 1
- 230000001684 chronic effect Effects 0.000 description 1
- 238000003759 clinical diagnosis Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 210000001061 forehead Anatomy 0.000 description 1
- 210000003780 hair follicle Anatomy 0.000 description 1
- 230000003054 hormonal effect Effects 0.000 description 1
- 230000028993 immune response Effects 0.000 description 1
- 239000000411 inducer Substances 0.000 description 1
- 150000002632 lipids Chemical class 0.000 description 1
- 206010025135 lupus erythematosus Diseases 0.000 description 1
- 244000005700 microbiome Species 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 239000011148 porous material Substances 0.000 description 1
- 230000035755 proliferation Effects 0.000 description 1
- 229940055019 propionibacterium acne Drugs 0.000 description 1
- 210000001732 sebaceous gland Anatomy 0.000 description 1
- 210000002374 sebum Anatomy 0.000 description 1
- 230000028327 secretion Effects 0.000 description 1
- 210000003491 skin Anatomy 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 210000000106 sweat gland Anatomy 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4053—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-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
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06T2207/10048—Infrared 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
- G06T2207/20081—Training; Learning
-
- 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
- G06T2207/20084—Artificial neural networks [ANN]
-
- 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
- G06T2207/20212—Image combination
- G06T2207/20221—Image fusion; Image merging
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Multimedia (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The invention discloses a correlation calculation method of a human face and a macaque face based on acne, belonging to the field of digital image processing and deep learning. The invention designs image processing, neural network, multi-mode fusion, contour similarity calculation, and data set production and acquisition. According to the invention, through expanding the data set, the multi-person face segmentation results are subjected to superposition fusion and contour fitting, the uniqueness of an individual is eliminated, and the problem of commonality is explored. And (5) calculating the similarity of the fitting contour to obtain a conclusion of whether the association exists between the acne occurrence area of the human face and the hair region of the macaque face.
Description
Technical Field
The invention belongs to the field of digital image processing and deep learning, and particularly relates to an automatic segmentation technology of human face acne and macaque bare surface area outline based on multi-mode fusion deep learning, and a correlation verification method of acne distribution and macaque face fuzzed area.
Background
The face is used as the first name card of people's daily communication, and the aesthetic degree directly influences the first impression in the mind of other people, and the importance is self-evident. The guide for treating acne in China (revised 2019) indicates that acne is a chronic skin inflammation which is good in puberty, more than 95% of people in China can generate acne with different degrees, part of patients can leave scars, negative psychology such as anxiety, pressure, depression and the like can be easily caused, and treatment methods need to be emphasized and standardized.
However, the pathogenesis of acne has not yet been fully elucidated, and its inducers include, but are not limited to, genetic factors, hormonal secretion, sebum hypersecretion, abnormal propionibacterium acnes reproduction, etc. It is currently widely accepted by clinical dermatology that the premise of acne is that sebaceous glands secrete large amounts of lipids, which lead to inflammation and immune responses caused by abnormal proliferation of hair follicle microorganisms.
The T area (the area formed by forehead and nose) of the face has more sweat gland pores and serious daily oil discharge; the skin of the U-shaped region (the left cheek region and the right cheek region) of the face is fine and smooth, and the oil output is small. Then, according to the pathogenesis, the T area is the acne frequent area, and the U area is the acne minor area. However, according to clinical diagnosis and treatment of acne patients, the occurrence probability of acne in the U area of severe acne patients is very high, and scars such as acne pits often appear in the U area. Meanwhile, after observing the acne occurrence area of a plurality of people, the outline of the acne-prone area is very similar to the outline of the macaque hair area. So that the outline of the frequent facial acne area is similar to the hair area of the macaque, which is suspected to be influenced by the genetic evolution of human beings.
Therefore, a new verification method is needed to solve the above problems, which can effectively verify the above guesses, supplement the pathogenesis of acne, establish the association with macaque, and enrich the follow-up experiment and diagnosis and treatment means.
Disclosure of Invention
Aiming at the guess that the facial acne distribution area of the acne patient is similar to the macaque face hair area in clinic at present, the invention designs a verification method combining an image color space and multi-mode deep learning, thereby achieving the purpose of objectively describing the relevance between the facial acne occurrence area and the macaque hair area.
The technical scheme of the invention is a correlation calculation method of human faces and macaque faces based on acne, which comprises the following steps:
step 1: acquiring face data; collecting a human face parallel polarization spectrum image and a near infrared spectrum shooting image by using a high-resolution camera;
step 2: screening the sample images acquired in the step 1, removing patient images with obvious skin diseases (such as vitiligo, lupus erythematosus and the like), and dividing a training set and a testing set according to the proportion of 1:9;
step 3: preprocessing the parallel polarization spectrum image processed in the step 2;
step 3-1: carrying out graying treatment on the parallel polarization spectrum image obtained in the step (2);
step 3-2: performing face key point calibration on the face information in the image by using OpenCV Dlib on the image processed in the step 3-1;
step 3-3: determining the contours of eyes, nose and mouth of the face by using the obtained face key points and OpenCV filter Poly functions processed in the step 3-2, and obtaining masks of the facial organs;
step 4: performing facial acne labeling on the near infrared spectrum image obtained in the step (2);
step 4-1: the facial organ mask obtained in the step 3-3 is overlapped on the near infrared spectrum image obtained in the step 1;
step 4-2: converting the superimposed mask image obtained in the step 4-1 into an image represented by HSV color space, and setting a threshold range according to the difference between the acne inflammation area and normal skin;
step 4-3: threshold segmentation is carried out on the image obtained in the step 4-2, a segmented image is obtained, the corresponding segmentation area is set to be 1, and the background area is set to be 0;
step 4-4: carrying out connected domain calculation on the image obtained in the step 4-3, and carrying out external rectangle on each connected domain;
step 4-5: converting the external rectangular data obtained in the step 4-4 into a JSON format file;
step 4-6: correcting the JSON file obtained in the step 4-5 and the parallel polarization spectrum image obtained in the step 2 by combining labeling software labelme, so as to ensure that no label leakage and false label exist;
step 5: processing the marked JSON file obtained in the step 4, and converting the marked JSON file into a VOC (volatile organic compound) format file;
step 6: fusing the parallel polarization spectrum image obtained in the step 1 with the near infrared spectrum image in the channel direction, and changing the image size from (2000,3000,3) to (2000,3000,6);
step 7: constructing a multi-mode convolutional neural network structure based on YOLOv5 adjustment;
step 8: the VOC format labeling file obtained in the step 5 and the fusion image obtained in the step 6 are sent to the multi-mode neural network built in the step 7 for training until the training is successful;
step 9: detecting the test set obtained in the step 2 by adopting the multi-mode convolutional neural network model which is trained in the step 8 and adjusted based on the YOLOv5, and additionally storing the acne area obtained by detection as a binary image, wherein the acne area is 1, and the background area is 0;
step 10: combining the binary images obtained in the step 9;
step 10-1: determining a face template M, and adjusting other images based on the face template M to ensure the facial five sense organs of all people to be basically aligned;
step 10-2: carrying out graying treatment on the parallel polarization spectrum image of the test set obtained in the step (2);
step 10-3: performing face key point calibration on the face information in the image by using OpenCV Dlib on the image processed in the step 10-2;
step 10-4: calculating the circumscribed rectangle centers O of all the feature points of the image processed in the step 10-3, and calculating the difference between the circumscribed rectangle centers O and the feature points of the template M;
step 10-5: calculating the included angle between the connecting line and the vertical line for the two feature points at the uppermost part and the lowermost part of the center axis of the connecting nose of the image processed in the step 10-3;
step 10-6: calculating the length and width of the circumscribed rectangle of all the feature points of the image processed in the step 10-3, and calculating the length ratio of the circumscribed rectangle of the feature points of the template M;
step 10-7: translating the binary image obtained in the step 9 according to the position difference obtained in the step 10-4, rotating the image according to the included angle obtained in the step 10-5, and scaling according to the aspect ratio obtained in the step 10-6;
step 10-8: superposing all the binary images processed in the step 10-7 to obtain a multi-person acne occurrence position diagram, so that the specificity is reduced;
step 11: performing contour fitting on the images obtained in the step 10-8 to obtain the internal contour of acne distribution;
step 12: obtaining macaque surface data;
step 13: labeling the monkey face data obtained in the step 12, and marking the boundary between the monkey face fur area and the bare surface area;
step 14: training the neural network model based on the U-net by adopting the file obtained in the step 13 until the training is successful;
step 15: dividing a rear test set by adopting the neural network model based on the U-net trained in the step 14 to obtain a monkey face bare surface area;
step 16: boundary extraction is carried out on all the monkey face bare surface areas obtained in the step 15, and the monkey face bare surface areas are fused into a boundary and filled, so that the specificity is reduced;
step 16-1: making external rectangles for all the monkey face bare surface areas obtained in the step 15;
step 16-2: defining a rectangular template, and determining the size of the rectangle and the degree of an included angle between the central axis and the coordinate axis;
step 16-3: performing rotation and scaling operations on the circumscribed rectangle and the internal image thereof obtained in the step 16-1, so that the circumscribed rectangle is consistent with the template defined in the step 16-2;
step 16-4: calculating all monkey face bare surface connected domains obtained in the step 16-3 respectively to obtain a central point O of the monkey face bare surface connected domains, simulating a polar coordinate system, taking O as an original point, taking 1 degree as an angle difference, and transmitting 360 rays to the periphery, wherein points where the rays intersect with the edges of the connected domains are sampling points;
step 16-5: taking the sampling points with the same corresponding angles obtained in the step 16-4 as a group of data, and respectively calculating and averaging each group of data to obtain 360 average value points;
step 16-6: smoothly connecting and filling the mean points obtained in the step 16-5 in sequence to obtain a monkey face bare surface area fusion map, and reducing the specificity;
step 17: and (3) carrying out similarity calculation on the distribution profile of the acne on the human face and the bare surface area of the monkey face, which are respectively obtained in the step (11) and the step (16), and respectively using three calculation methods of Euclidean distance, cosine distance and Haosdorf distance to calculate, wherein the obtained result is used for objectively evaluating the similarity degree of the profiles of the two.
According to the invention, the multi-modal neural network is adopted to extract the acne features fusing the parallel polarization spectrum image and the near infrared spectrum image, the recognition precision is higher, the means of image rotation, translation and scaling are adopted to enable the five sense organs and cheeks of different people to be distributed in the same area as much as possible, the method of image fusion and sample size increase is adopted to reduce the difference between different individuals, and a plurality of evaluation parameters are used for calculation, so that the similarity evaluation of the two is more objective and accurate.
Drawings
FIG. 1 is an overall workflow of the present invention.
Fig. 2 is a convolutional neural network flow for intelligent monitoring of facial acne in accordance with the present invention.
Fig. 3 is a flow chart of facial acne dataset production in accordance with the present invention.
FIG. 4 is a flow chart of the monkey face contour fitting of the present invention.
Detailed Description
In order to make the objects, technical solutions and innovative features of the present invention more apparent, the present invention will be described in more detail with reference to the accompanying drawings and embodiments, it being understood that the embodiments described herein are for illustrative purposes only and are not limiting of the invention.
Referring to fig. 1 to 4, the invention relates to a method for exploring the association between a facial acne occurrence area and a macaque hair area based on an image color and convolutional neural network, which comprises an automatic segmentation and extraction module for human facial acne, and is used for automatically and intelligently segmenting and extracting a plurality of human facial acnes; the macaque face bare surface area segmentation module is used for segmenting and extracting macaque face areas from a plurality of macaque face images; the image merging contour fitting module is used for respectively carrying out contour fitting on the facial acne segmentation image and the macaque nude segmentation image; and the similarity calculation module is used for calculating the similarity degree of the fitting contour based on the acne and the fitting contour of the macaque surface.
A method for calculating the relevance between a facial acne occurrence area and a macaque hair area based on an image color and convolutional neural network comprises the following steps:
step 1-1: acquiring face data, namely acquiring a multi-face parallel polarization spectrum image and a near infrared spectrum shooting image by using two different light sources respectively by using a high-resolution camera capable of fixing faces;
step 1-2: screening the sample images acquired in the step 1-1, and dividing a training set and a testing set according to the proportion of 1:9;
step 1-3: carrying out graying treatment on the parallel polarization spectrum image obtained in the step 1-2, carrying out face key point calibration, and obtaining masks of eyes, nose and mouth of the face according to the key points;
step 1-4: after the near infrared spectrum image is overlapped with a mask, converting the whole image into an image represented by HSV color space by RGB representation, and setting a threshold range to extract acne preliminarily;
step 1-5: performing manual correction on the basis of the coarse mark to ensure that no missing mark or false mark exists, and manufacturing the coarse mark into a VOC (volatile organic compound) format file;
step 1-6: training the input data by utilizing a multi-mode convolutional neural network structure based on YOLOv5 adjustment until the loss drop tends to be stable;
step 1-7: detecting the test set obtained in the step 1-2 by adopting the network model trained in the step 1-6, and additionally storing the acne area obtained by detection as a binary image;
step 1-8: combining the binary images obtained in the steps 1-7, carrying out translation, rotation and scaling on other images according to a template to enable all faces to be basically aligned, and then superposing the binary images obtained by model segmentation to obtain a multi-person acne occurrence position diagram, so that the specificity is reduced;
step 2-1: obtaining monkey face data and marking a monkey face bare surface area;
step 2-2: training the input macaque face data by using a neural network model based on U-net until the training is successful;
step 2-3: dividing a large number of macaque face images by adopting the network model trained in the step 2-2 to obtain a macaque face bare surface area;
step 2-4: carrying out connected domain calculation on all monkey face bare surface areas obtained in the step 2-3, making external rectangles, and rotating and scaling according to a defined rectangular template until all monkey face connected domain external rectangles are consistent;
step 2-5: calculating the monkey face connected domain processed in the step 2-4 to obtain a central point O, simulating a polar coordinate system, taking O as an original point, taking 1 degree as an angle interval, and taking rays to the periphery, wherein the point where the rays intersect with the edge of the connected domain is the sampling point of the angle;
step 2-6: averaging the sampling points corresponding to the same angle obtained in the step 2-5 to obtain 360 average value points, sequentially smoothly linking and filling to obtain a fitted monkey face bare surface area, and reducing the specificity;
step 3: and (3) carrying out similarity calculation on the distribution profile of the acne on the human face and the bare surface area of the monkey face, which are respectively obtained in the steps (1-8) and (2-6), and respectively using three calculation methods of Euclidean distance, cosine distance and Haosdorf distance to calculate, wherein the obtained result is used for objectively evaluating the similarity degree of the profiles of the two.
The invention automatically identifies the characteristics of human acne and the characteristics of the bare surface-fuzzed boundary of the macaque face by using a deep learning network, thereby realizing the possibility of batch processing of data and reducing the subjectivity of manual division; increasing the sample size, adopting a proper method to respectively perform image fusion on the human face and the monkey face, and the like to reduce the difference between different individuals; and a plurality of evaluation parameters are used for calculation, so that the similarity evaluation of the two parameters is more objective and accurate.
Claims (2)
1. A method of calculating the association of a human face with a macaque face based on acne, the method comprising:
step 1: acquiring face data; collecting a human face parallel polarization spectrum image and a near infrared spectrum shooting image by using a high-resolution camera;
step 2: screening the sample images acquired in the step 1, removing the patient images with obvious skin diseases, and dividing a training set and a testing set according to the proportion of 1:9;
step 3: preprocessing the parallel polarization spectrum image processed in the step 2;
step 3-1: carrying out graying treatment on the parallel polarization spectrum image obtained in the step (2);
step 3-2: performing face key point calibration on the face information in the image by using OpenCV Dlib on the image processed in the step 3-1;
step 3-3: determining the contours of eyes, nose and mouth of the face by using the obtained face key points and OpenCV filter Poly functions processed in the step 3-2, and obtaining masks of the facial organs;
step 4: performing facial acne labeling on the near infrared spectrum image obtained in the step (2);
step 4-1: the facial organ mask obtained in the step 3-3 is overlapped on the near infrared spectrum image obtained in the step 1;
step 4-2: converting the superimposed mask image obtained in the step 4-1 into an image represented by HSV color space, and setting a threshold range according to the difference between the acne inflammation area and normal skin;
step 4-3: threshold segmentation is carried out on the image obtained in the step 4-2, a segmented image is obtained, the corresponding segmentation area is set to be 1, and the background area is set to be 0;
step 4-4: carrying out connected domain calculation on the image obtained in the step 4-3, and carrying out external rectangle on each connected domain;
step 4-5: converting the external rectangular data obtained in the step 4-4 into a JSON format file;
step 4-6: correcting the JSON file obtained in the step 4-5 and the parallel polarization spectrum image obtained in the step 2 by combining labeling software labelme, so as to ensure that no label leakage and false label exist;
step 5: processing the marked JSON file obtained in the step 4, and converting the marked JSON file into a VOC (volatile organic compound) format file;
step 6: fusing the parallel polarization spectrum image obtained in the step 1 with the near infrared spectrum image in the channel direction, and changing the image size from (2000,3000,3) to (2000,3000,6);
step 7: constructing a multi-mode convolutional neural network structure based on YOLOv5 adjustment;
step 8: the VOC format labeling file obtained in the step 5 and the fusion image obtained in the step 6 are sent to the multi-mode neural network built in the step 7 for training until the training is successful;
step 9: detecting the test set obtained in the step 2 by adopting the multi-mode convolutional neural network model which is trained in the step 8 and adjusted based on the YOLOv5, and additionally storing the acne area obtained by detection as a binary image, wherein the acne area is 1, and the background area is 0;
step 10: combining the binary images obtained in the step 9;
step 10-1: determining a face template M, and adjusting other images on a sequential basis to ensure the facial five sense organs of all people to be basically aligned;
step 10-2: carrying out graying treatment on the parallel polarization spectrum image of the test set obtained in the step (2);
step 10-3: performing face key point calibration on the face information in the image by using OpenCV Dlib on the image processed in the step 10-2;
step 10-4: calculating the circumscribed rectangle centers O of all the feature points of the image processed in the step 10-3, and calculating the difference between the circumscribed rectangle centers O and the feature points of the template M;
step 10-5: calculating the included angle between the connecting line and the vertical line for the two feature points at the uppermost part and the lowermost part of the center axis of the connecting nose of the image processed in the step 10-3;
step 10-6: calculating the length and width of the circumscribed rectangle of all the feature points of the image processed in the step 10-3, and calculating the length ratio of the circumscribed rectangle of the feature points of the template M;
step 10-7: translating the binary image obtained in the step 9 according to the position difference obtained in the step 10-4, rotating the image according to the included angle obtained in the step 10-5, and scaling according to the aspect ratio obtained in the step 10-6;
step 10-8: superposing all the binary images processed in the step 10-7 to obtain a multi-person acne occurrence position diagram, so that the specificity is reduced;
step 11: performing contour fitting on the images obtained in the step 10-8 to obtain the internal contour of acne distribution;
step 12: obtaining macaque surface data;
step 13: labeling the monkey face data obtained in the step 12, and marking the boundary between the monkey face fur area and the bare surface area;
step 14: training the neural network model based on the U-net by adopting the file obtained in the step 13 until the training is successful;
step 15: dividing a rear test set by adopting the neural network model based on the U-net trained in the step 14 to obtain a monkey face bare surface area;
step 16: boundary extraction is carried out on all the monkey face bare surface areas obtained in the step 15, and the monkey face bare surface areas are fused into a boundary and filled, so that the specificity is reduced;
step 17: and (3) carrying out similarity calculation on the distribution profile of the acne on the human face and the bare surface area of the monkey face, which are respectively obtained in the step (11) and the step (16), and respectively using three calculation methods of Euclidean distance, cosine distance and Haosdorf distance to calculate, wherein the obtained result is used for objectively evaluating the similarity degree of the profiles of the two.
2. A method for calculating the association between the face of a person and the face of a macaque based on acne according to claim 1, wherein the specific method of step 16 is as follows: step 16-1: making external rectangles for all the monkey face bare surface areas obtained in the step 15;
step 16-2: defining a rectangular template, and determining the size of the rectangle and the degree of an included angle between the central axis and the coordinate axis;
step 16-3: performing rotation and scaling operations on the circumscribed rectangle and the internal image thereof obtained in the step 16-1, so that the circumscribed rectangle is consistent with the template defined in the step 16-2;
step 16-4: calculating all monkey face bare surface connected domains obtained in the step 16-3 respectively to obtain a central point O of the monkey face bare surface connected domains, simulating a polar coordinate system, taking O as an original point, taking 1 degree as an angle difference, and transmitting 360 rays to the periphery, wherein points where the rays intersect with the edges of the connected domains are sampling points;
step 16-5: taking the sampling points with the same corresponding angles obtained in the step 16-4 as a group of data, and respectively calculating and averaging each group of data to obtain 360 average value points;
step 16-6: and (3) sequentially and smoothly connecting and filling the mean value points obtained in the step (16-5) to obtain a monkey face bare surface area fusion map, and reducing the specificity.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311127893.1A CN117274673A (en) | 2023-09-01 | 2023-09-01 | Correlation calculation method of human face and macaque face based on acne |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311127893.1A CN117274673A (en) | 2023-09-01 | 2023-09-01 | Correlation calculation method of human face and macaque face based on acne |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117274673A true CN117274673A (en) | 2023-12-22 |
Family
ID=89215097
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311127893.1A Pending CN117274673A (en) | 2023-09-01 | 2023-09-01 | Correlation calculation method of human face and macaque face based on acne |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117274673A (en) |
-
2023
- 2023-09-01 CN CN202311127893.1A patent/CN117274673A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110619301B (en) | Emotion automatic identification method based on bimodal signals | |
| CN112967285A (en) | Chloasma image recognition method, system and device based on deep learning | |
| CN112070772A (en) | Blood leukocyte image segmentation method based on UNet + + and ResNet | |
| CN110458831B (en) | Scoliosis image processing method based on deep learning | |
| CN111666845B (en) | Small sample deep learning multi-mode sign language recognition method based on key frame sampling | |
| WO2011074014A2 (en) | A system for lip corner detection using vision based approach | |
| CN108615239A (en) | Tongue image dividing method based on threshold technology and Gray Projection | |
| Monwar et al. | Pain recognition using artificial neural network | |
| CN106778047A (en) | A kind of traditional Chinese medical science facial diagnosis integrated system based on various dimensions medical image | |
| CN104933723B (en) | Tongue image dividing method based on rarefaction representation | |
| Xue et al. | Optic disk detection and segmentation for retinal images using saliency model based on clustering | |
| CN109711306B (en) | Method and equipment for obtaining facial features based on deep convolutional neural network | |
| CN117274673A (en) | Correlation calculation method of human face and macaque face based on acne | |
| CN117809030A (en) | Breast cancer CT image identification and segmentation method based on artificial neural network | |
| CN108629780B (en) | Tongue image segmentation method based on color decomposition and threshold technology | |
| CN111814738A (en) | Human face recognition method, human face recognition device, computer equipment and medium based on artificial intelligence | |
| CN113160224B (en) | Artificial intelligence-based skin aging degree identification method, system and device | |
| CN115601339A (en) | Tongue picture inquiry moxibustion scheme generation system and method based on artificial intelligence | |
| Lin et al. | Glandular Cell Image Segmentation Method based on Improved SegNet Neural Network | |
| CN110751661A (en) | Clustering algorithm-based facial chloasma region automatic segmentation method | |
| Rani et al. | Automatic Segmentation and Classification of Skin Cancer Cells using Thresholding and Deep Learning based Techniques | |
| CN116258697B (en) | Automatic classification device and method for child skin disease images based on rough labeling | |
| Chen et al. | The development of a skin inspection imaging system on an Android device | |
| Xin et al. | A fast tongue image color correction method based on gray world method | |
| CN111914632B (en) | Face recognition method, device and storage medium |
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 |