CN113344095B - K-means worker fatigue simple discrimination method based on multi-feature operator - Google Patents
K-means worker fatigue simple discrimination method based on multi-feature operator Download PDFInfo
- Publication number
- CN113344095B CN113344095B CN202110686803.7A CN202110686803A CN113344095B CN 113344095 B CN113344095 B CN 113344095B CN 202110686803 A CN202110686803 A CN 202110686803A CN 113344095 B CN113344095 B CN 113344095B
- Authority
- CN
- China
- Prior art keywords
- image
- training sample
- characterization
- fatigue
- images
- 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
- 238000012850 discrimination method Methods 0.000 title claims description 5
- 238000012549 training Methods 0.000 claims abstract description 108
- 238000000605 extraction Methods 0.000 claims abstract description 54
- 238000000034 method Methods 0.000 claims abstract description 32
- 238000005457 optimization Methods 0.000 claims abstract description 25
- 238000012545 processing Methods 0.000 claims abstract description 22
- 201000011243 gastrointestinal stromal tumor Diseases 0.000 claims abstract description 18
- 238000003064 k means clustering Methods 0.000 claims abstract description 14
- 238000012512 characterization method Methods 0.000 claims description 64
- 108010063499 Sigma Factor Proteins 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 abstract description 4
- 230000001815 facial effect Effects 0.000 description 5
- 238000005286 illumination Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 239000003086 colorant Substances 0.000 description 3
- 208000020154 Acnes Diseases 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000037303 wrinkles Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 230000003014 reinforcing effect Effects 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/254—Fusion techniques of classification results, e.g. of results related to same input data
- G06F18/256—Fusion techniques of classification results, e.g. of results related to same input data of results relating to different input data, e.g. multimodal recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Probability & Statistics with Applications (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a simple and easy judging method of K-means workers fatigue based on a multi-feature operator, and relates to the technical field of fatigue judgment; performing SIFT feature extraction, GIST feature extraction and HOG feature extraction on the positive training sample image and the negative training sample image which are subjected to optimization treatment respectively; judging whether the image to be detected is a fatigue picture or a non-fatigue picture or not by using a K-means clustering algorithm; the method is fully combined by the methods of multi-scale image detail optimization processing, multi-feature operator feature extraction, K-means algorithm and the like, so that fatigue of workers is judged more comprehensively, accurately and robustly; the model used by the whole system is simpler, has no larger consumption of calculation resources, greatly saves related resources and reduces operation time.
Description
Technical Field
The invention relates to the technical field of computers, in particular to a simple and easy judging method for fatigue of K-means workers based on a multi-feature operator.
Background
In the current society, workers often bear huge working pressure. In some cases, workers need to work over to complete tasks, and fatigue is very likely to occur. When the worker is tired, obvious working errors are easy to occur, and the health of the person can be endangered when serious. More importantly, long-term fatigue tends to cause very serious diseases. In addition, due to regulatory, security, etc., modern equipment of a factory can often capture images of workers relatively easily. Therefore, how to make fatigue judgment by using the image of the worker is a very significant task.
Aiming at the problems, in order to ensure higher precision, the traditional method not only depends on a plurality of images, videos and the like, but also often uses a complex training and distinguishing model to train and make decisions. Even if higher discrimination accuracy is obtained, more computing resources are often consumed. Therefore, how to establish a worker fatigue judging method based on a single image and judge whether the worker is tired or not by using a model with lower consumption is a very significant and urgent work to be solved.
Disclosure of Invention
In order to overcome the problems or at least partially solve the problems, the embodiment of the invention provides a simple method for judging the fatigue of K-means workers based on multi-feature operators, which judges whether the workers are tired or not by using a model with lower consumption.
Embodiments of the present invention are implemented as follows:
a simple and easy discriminating method of K-means workman fatigue based on multi-feature operator includes the following steps:
selecting face images of a plurality of non-tired workers as positive training sample images;
selecting face images of a plurality of tired workers as negative training sample images;
respectively carrying out multi-scale image detail optimization processing on the selected positive training sample image and negative training sample image to obtain an optimized positive training sample image and negative training sample image;
respectively carrying out SIFT feature extraction on the positive training sample image and the negative training sample image after optimization processing, and carrying out characterization on the image to be detected by using a SIFT feature operator to obtain a first characterization image;
judging the first characterization image by using a K-means clustering algorithm to obtain a first judging result, wherein the first judging result comprises a fatigue picture or a non-fatigue picture;
performing GIST feature extraction on the positive training sample image and the negative training sample image which are subjected to optimization processing respectively, and utilizing a GIST feature operator to realize representation of the image to be detected so as to obtain a second representation image;
judging the second characterization image by using a K-means clustering algorithm to obtain a second judging result, wherein the second judging result comprises a fatigue picture or a non-fatigue picture;
performing HOG feature extraction on the positive training sample image and the negative training sample image which are subjected to optimization processing respectively, and performing characterization on the image to be detected by using an HOG feature operator to obtain a third characterization image;
judging the third characterization image by using a K-means clustering algorithm to obtain a third judging result, wherein the third judging result comprises a fatigue picture or a non-fatigue picture;
if two or more of the first, second and third discrimination results of the image to be detected are judged to be non-fatigue pictures, the image to be detected is finally judged to be the non-fatigue pictures;
if two or more of the first, second and third judging results of the image to be detected are judged to be fatigue pictures, the image to be detected is finally judged to be the fatigue pictures.
The method is fully combined by the methods of multi-scale image detail optimization processing, multi-feature operator feature extraction, K-means algorithm and the like, so that fatigue of workers is judged more comprehensively, accurately and robustly; the model used by the whole system is simpler, has no larger consumption of calculation resources, greatly saves related resources and reduces operation time.
In some embodiments of the present invention, the multi-scale image detail optimization processing for the selected positive training sample image and the negative training sample image respectively includes the following steps:
filtering the image using a least squares filter;
dividing the remaining positive training sample image and negative training sample image into a plurality of scales;
performing subtraction computation among different scales to obtain detail information with different degrees;
weighting detail information with different degrees into an original image;
and obtaining the enhanced image containing abundant detail information.
In some embodiments of the present invention, discriminating the first characterization image, the second characterization image, and the three characterization images, respectively, using a K-means clustering algorithm includes the steps of:
randomly selecting 2 initial clustering centers from the first characterization image, the second characterization image, the third characterization image, the plurality of positive training sample images and the plurality of negative training sample images;
calculating the distance between each sample and each cluster center, and classifying each sample to the cluster center closest to the sample;
for each cluster, taking the average value of all samples as a new cluster center of the cluster;
repeating the steps until the clustering center is not changed;
if the images to be detected are clustered into positive sample types, judging the images to be non-fatigue images;
if the images to be detected are clustered into a negative sample class, we determine them as fatigue images.
In some embodiments of the present invention, SIFT feature extraction and use SIFT feature operators to implement characterization of images, GIST feature extraction and use GIST feature operators to implement characterization of images and HOG feature extraction and use HOG feature operators to implement characterization of images running side by side.
In some embodiments of the invention, SIFT feature extraction comprises:
constructing a scale space; the scale space of the defined image is:
；
wherein G is a gaussian function:
；
wherein G (x, y, sigma) is a scale variable Gaussian function, (x, y) is a space coordinate, sigma is a scale space factor, sigma is a standard deviation of Gaussian normal distribution, reflects the degree of blurring of an image, and the larger the value is, the more blurring of the image is, the larger the corresponding scale is, and L (x, y, sigma) corresponds to Gaussian scale space.
In some embodiments of the present invention, the step of putting the characterization image, the plurality of positive training sample images, and the plurality of negative training sample images together, the number of positive training samples and the number of negative training samples are identical.
In some embodiments of the invention, the number of positive training samples and negative training samples is at least greater than 50, respectively.
In some embodiments of the present invention, the number of positive training samples and negative training samples is 100, respectively.
In some embodiments of the invention, the method of selecting comprises: and (5) manual selection.
In some embodiments of the invention, the method of selecting comprises: and (5) selecting a machine.
The embodiment of the invention has at least the following advantages or beneficial effects:
the invention provides a simple and easy judging method of K-means workman fatigue based on multi-feature operator, through choosing the facial image of a plurality of non-tired workman as the positive training sample image; selecting face images of a plurality of tired workers as negative training sample images; respectively carrying out multi-scale image detail optimization processing on the selected positive training sample image and negative training sample image to obtain an optimized positive training sample image and negative training sample image; respectively carrying out SIFT feature extraction on the positive training sample image and the negative training sample image after optimization processing, and carrying out characterization on the image to be detected by using a SIFT feature operator to obtain a first characterization image; judging the first characterization image by using a K-means clustering algorithm to obtain a first judging result, wherein the first judging result comprises a fatigue picture or a non-fatigue picture; performing GIST feature extraction on the positive training sample image and the negative training sample image which are subjected to optimization processing respectively, and utilizing a GIST feature operator to realize representation of the image to be detected so as to obtain a second representation image; judging the second characterization image by using a K-means clustering algorithm to obtain a second judging result, wherein the second judging result comprises a fatigue picture or a non-fatigue picture; performing HOG feature extraction on the positive training sample image and the negative training sample image which are subjected to optimization processing respectively, and performing characterization on the image to be detected by using an HOG feature operator to obtain a third characterization image; judging the third characterization image by using a K-means clustering algorithm to obtain a third judging result, wherein the third judging result comprises a fatigue picture or a non-fatigue picture; if two or more of the first, second and third discrimination results of the image to be detected are judged to be non-fatigue pictures, the image to be detected is finally judged to be the non-fatigue pictures; if two or more of the first, second and third judging results of the image to be detected are judged to be fatigue pictures, the image to be detected is finally judged to be the fatigue pictures.
The method is fully combined by the methods of multi-scale image detail optimization processing, multi-feature operator feature extraction, K-means algorithm and the like, so that fatigue of workers is judged more comprehensively, accurately and robustly; the model used by the whole system is simpler, has no larger consumption of calculation resources, greatly saves related resources and reduces operation time.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the embodiments will be briefly described below, it being understood that the following drawings only illustrate some embodiments of the present invention and therefore should not be considered as limiting the scope, and other related drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of an embodiment of a simple method for judging fatigue of K-means workers based on a multi-feature operator;
FIG. 2 is a flowchart of another embodiment of a simple method for judging fatigue of K-means workers based on a multi-feature operator according to the present invention;
FIG. 3 is a flowchart of another embodiment of a simple method for judging fatigue of K-means workers based on multi-feature operators.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. The components of the embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.
Thus, the following detailed description of the embodiments of the invention, as presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Examples
Referring to fig. 1-3, the embodiment provides a simple method for judging fatigue of K-means workers based on multi-feature operators, which comprises the following steps:
s1: selecting face images of a plurality of non-tired workers as positive training sample images, and selecting face images of a plurality of tired workers as negative training sample images;
the method for selecting the face images of the fatigue workers can be used for manually selecting the face images of the fatigue workers as positive training sample images and the face images of the non-fatigue workers as negative training samples; the machine may also select a plurality of facial images of non-tired workers as positive training sample images and facial images of tired workers as negative training samples. The number of positive training samples and negative training samples can be selected according to actual conditions, in the experiment, the number of positive training samples is equal to the number of negative training samples, the number of positive training samples and the number of negative training samples are respectively at least greater than 50, if the number of positive training samples and the number of negative training samples are less than 50, the result is inaccurate, and the number of positive training samples and the number of negative training samples are respectively 100.
S2: respectively carrying out multi-scale image detail optimization processing on the selected positive training sample image and negative training sample image to obtain an optimized positive training sample image and negative training sample image;
for example, referring to fig. 2, the steps of performing multi-scale image detail optimization on the selected positive training sample image and the negative training sample image respectively include:
s21: filtering the image using a least squares filter;
s22: dividing the remaining positive training sample image and negative training sample image into a plurality of scales;
s23: performing subtraction computation among different scales to obtain detail information with different degrees;
s24: weighting detail information with different degrees into an original image;
s25: and obtaining the enhanced image containing abundant detail information.
Similar to Retinex, three-scale Gaussian blur is used, subtraction is carried out on the three-scale Gaussian blur and the original image, so that detail information with different degrees is obtained, and then the detail information is fused into the original image in a certain combination mode, so that the capability of reinforcing the original image information is obtained.
The programming of the algorithm in the multi-scale image detail optimization process comprises the following steps:
Int IM_MultiScaleSharpen(unsigned char*Src，unsigned char *Dest， int Width，int Height，int Stride，int Radius)
{
Int Channel=Stride/Width；
if((Src==NULL)||(Dest==NULL)) returnIM_STATUS_NULLREFRENCE；
if((Width<=0)||(Height<=0)) returnIM_STATUS_INVALIDPARAMETER；
if((Channel!=1)&&(Channel!=3)&&(Channel!=4)) returnIM_STATUS_INVALIDPARAMETER；
intStatus=IM_STATUS_OK；
unsignedchar*B1=(unsignedchar*)malloc(Height*Stride*sizeof(unsignedchar))；
unsignedchar*B2=(unsignedchar*)malloc(Height*Stride*sizeof(unsignedchar))；
unsignedchar*B3=(unsignedchar*)malloc(Height*Stride*sizeof(unsignedchar))；
if((B1==NULL)||(B2==NULL)||(B3==NULL))
{
if(B1!=NULL)free(B1);
if(B2!=NULL)free(B2);
if(B3!=NULL)free(B3);
returnIM_STATUS_OUTOFMEMORY；
}
Status=IM_ExpBlur(Src,B1,Width,Height,Stride,Radius)；
if(Status!=IM_STATUS_OK)gotoFreeMemory；
Status=IM_ExpBlur(Src，B2，Width,Height,Stride,Radius*2)；
if(Status!=IM_STATUS_OK)gotoFreeMemory；
Status=IM_ExpBlur(Src，B3，Width，Height，Stride，Radius*4)；
if(Status!=IM_STATUS_OK)gotoFreeMemory；
for(intY=0;Y<Height*Stride；Y++)
{
intDiffB1=Src[Y]-B1[Y]；
intDiffB2=B1[Y]-B2[Y]；
intDiffB3=B2[Y]-B3[Y]；
Dest[Y]=IM_ClampToByte(((4-2*IM_Sign(DiffB1))*DiffB1+2*DiffB2+DiffB3)/4+Src[Y])；
}
FreeMemory:
free(B1)；
free(B2)；
free(B3)；
returnStatus；
}
s3: respectively carrying out SIFT feature extraction on the positive training sample image and the negative training sample image after optimization processing, and carrying out characterization on the image to be detected by using a SIFT feature operator to obtain a first characterization image;
wherein G (x, y, sigma) is a scale variable Gaussian function, (x, y) is a space coordinate, sigma is a scale space factor, sigma is a standard deviation of Gaussian normal distribution, reflects the degree of blurring of an image, and the larger the value is, the more blurring of the image is, the larger the corresponding scale is, and L (x, y, sigma) corresponds to Gaussian scale space. The sigma magnitude determines the smoothness of the image, the large scale corresponds to the profile features of the image, the small scale corresponds to the detail features of the image, and the large sigma value corresponds to the coarse scale, and vice versa, corresponds to the fine scale.
SIFT feature extraction is robust to rotation, scaling, brightness and the like, and is a very stable local feature, so that the SIFT feature extraction is adopted, the position change of the face of a worker is generated, and the face of the worker can be detected in a highlight state.
S3-1: judging the first characterization image by using a K-means clustering algorithm to obtain a first judging result, wherein the first judging result comprises a fatigue picture or a non-fatigue picture;
for example, referring to fig. 3, the distinguishing the first characterization image, the second characterization image and the three characterization images by using the K-means clustering algorithm includes the following steps:
s31, randomly selecting 2 initial clustering centers from the first characterization image, the second characterization image, the third characterization image, the plurality of positive training sample images and the plurality of negative training sample images;
s32, calculating the distance between each sample and each cluster center, and classifying each sample to the cluster center closest to the sample;
s33, taking the average value of all samples as a new cluster center of each cluster;
s34, repeating the steps until the clustering center is not changed;
s35, if the image to be detected is clustered into a positive sample type, judging the image to be a non-fatigue image;
if the images to be detected are clustered into the negative sample type, S36, the images to be detected are judged to be fatigue images.
S4: performing GIST feature extraction on the positive training sample image and the negative training sample image which are subjected to optimization processing respectively, and utilizing a GIST feature operator to realize representation of the image to be detected so as to obtain a second representation image;
the GIST feature extraction is specially used for extracting the information of the picture to remove redundancy, if the content of the picture is complex, the GIST feature extraction can effectively extract useful feature information, reduce the data volume, and the extracted features at the moment obviously increase the distinction degree of the picture compared with the original information. Therefore, GIST feature extraction is performed specifically on images with complicated faces to effectively extract useful feature information, such as wrinkles, pits and acnes on the faces, and the images can be accurately determined as fatigue images or non-fatigue images.
S4-1: judging the second characterization image by using a K-means clustering algorithm to obtain a second judging result, wherein the second judging result comprises a fatigue picture or a non-fatigue picture;
similarly, if the images to be detected are clustered into positive sample types, judging the images to be non-fatigue images; if the images to be detected are clustered into a negative sample class, we determine them as fatigue images.
S5: performing HOG feature extraction on the positive training sample image and the negative training sample image which are subjected to optimization processing respectively, and performing characterization on the image to be detected by using an HOG feature operator to obtain a third characterization image;
it should be noted that, both HOG feature extraction and SIFT feature extraction are descriptors, and the HOG feature extraction is mainly different from SIFT feature extraction:
(1) SIFT feature extraction is based on the description of keypoint feature vectors.
(2) HOG feature extraction is to uniformly divide the image into adjacent patches and then to make a histogram of gradients across all patches.
(3) SIFT feature extraction requires extreme points to pixels in image scale space, but not in HOG.
(4) SIFT feature extraction generally has two major steps, the first step is to extract feature points for an image, while HOG does not.
The invention has the advantages of using HOG feature extraction:
(1) HOG feature extraction represents structural features of edges (gradients), so local shape information can be described;
(2) The quantization of the position and direction space can inhibit the influence caused by translation and rotation to a certain extent;
(3) The histogram is normalized in the local area, so that the influence caused by illumination change can be partially offset;
(4) The influence of illumination colors on the image is ignored to a certain extent, so that the dimension of characterization data required by the image is reduced;
(5) Moreover, due to the processing method of the blocking and dividing units, the relation between the local pixel points of the image can be well characterized.
The HOG feature extraction can just describe local shape information, so that the influence of illumination colors on the image is ignored to a certain extent, and the relationship between local pixel points of the image can be well characterized.
S5-1: judging the third characterization image by using a K-means clustering algorithm to obtain a third judging result, wherein the third judging result comprises a fatigue picture or a non-fatigue picture;
similarly, if the images to be detected are clustered into positive sample types, judging the images to be non-fatigue images; if the images to be detected are clustered into a negative sample class, we determine them as fatigue images.
S6: if two or more of the first, second and third discrimination results of the image to be detected are judged to be non-fatigue pictures, the image to be detected is finally judged to be the non-fatigue pictures; if two or more of the first, second and third judging results of the image to be detected are judged to be fatigue pictures, the image to be detected is finally judged to be the fatigue pictures.
Specifically, SIFT feature extraction is robust to rotation, scaling, brightness, and the like, and is a very stable local feature, so that position change of the face of a worker can be detected even in a highlight state by adopting SIFT feature extraction.
The GIST feature extraction is specially used for effectively extracting useful feature information from a facial complex image, such as facial wrinkles, pits and acnes, and can accurately judge whether the images are fatigue images or non-fatigue images.
The HOG feature extraction can just describe local shape information, so that the influence of illumination colors on the image is ignored to a certain extent, and the relationship between local pixel points of the image can be well characterized.
It should be noted that, in order to ensure higher accuracy, the conventional method not only depends on multiple images, videos, and the like, but also uses a complex training and distinguishing model to train and make decisions. Even if higher discrimination accuracy is obtained, more computing resources are often consumed. While the feature extraction can judge the fatigue of workers, the single feature extraction has errors, and the accuracy of the judgment is required to be improved. In view of this, this application adopts SIFT feature extraction, GIST feature extraction and HOG feature extraction respectively to judge whether the worker is tired, and each feature extraction all has pertinence to the image discrimination, has avoided leaking the characteristic. Compared with single feature extraction, the method has the advantages that fatigue of workers is judged more comprehensively, accurately and robustly, and judgment accuracy is greatly improved.
The fatigue of workers is judged more comprehensively, accurately and robustly by fully combining the methods such as multi-scale image detail optimization processing, multi-feature operator feature extraction, K-means algorithm and the like; the model used by the whole system is simpler, has no larger consumption of calculation resources, greatly saves related resources and reduces operation time.
In some embodiments of the present invention, step S3, step S4, step S5 may run in parallel.
The above is only a preferred embodiment of the present invention, and is not intended to limit the present invention, but various modifications and variations can be made to the present invention by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
It will be evident to those skilled in the art that the present application is not limited to the details of the foregoing illustrative embodiments, and that the present application may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the application being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned.
Claims (9)
1. A simple and easy discriminating method of K-means workman fatigue based on multi-feature operator is characterized by comprising the following steps:
selecting face images of a plurality of non-tired workers as positive training sample images;
selecting face images of a plurality of tired workers as negative training sample images;
respectively carrying out multi-scale image detail optimization processing on the selected positive training sample image and negative training sample image to obtain an optimized positive training sample image and negative training sample image;
respectively carrying out SIFT feature extraction on the positive training sample image and the negative training sample image after optimization processing, and carrying out characterization on the image to be detected by using a SIFT feature operator to obtain a first characterization image;
performing GIST feature extraction on the positive training sample image and the negative training sample image which are subjected to optimization processing respectively, and utilizing a GIST feature operator to realize representation of the image to be detected so as to obtain a second representation image;
performing HOG feature extraction on the positive training sample image and the negative training sample image which are subjected to optimization processing respectively, and performing characterization on the image to be detected by using an HOG feature operator to obtain a third characterization image;
respectively judging the first characterization image, the second characterization image and the third characterization image by using a K-means clustering algorithm to obtain three judging results;
if two or more of the three judging results are judged to be the non-fatigue pictures, finally judging the three judging results to be the non-fatigue pictures;
if two or more of the three judging results are judged to be fatigue pictures, finally judging the three judging results to be fatigue pictures;
the method for distinguishing the first characterization image, the second characterization image and the third characterization image by using the K-means clustering algorithm comprises the following steps of:
putting the characterization image, the plurality of positive training sample images and the plurality of negative training sample images together, and randomly selecting 2 initial clustering centers from the characterization image, the plurality of positive training sample images and the plurality of negative training sample images; the characterization images are first characterization images, second characterization images or third characterization images;
calculating the distance between each characterization image, a plurality of positive training sample images and a plurality of negative training sample images and each cluster center, and attributing each characterization image, a plurality of positive training sample images and a plurality of negative training sample images to the cluster center closest to the characterization image, the positive training sample images and the negative training sample images;
for each cluster, taking the mean values of all the characterization images, the positive training sample images and the negative training sample images as new cluster centers of the clusters;
repeating the steps until the clustering center is not changed;
if the characterization images are clustered into positive sample categories, judging the positive sample categories as non-fatigue images;
if the characterization images are clustered into negative categories, they are determined to be fatigue images.
2. The simple discrimination method for fatigue of K-means workers based on multi-feature operators as claimed in claim 1, wherein the multi-scale image detail optimization processing for the selected positive training sample image and negative training sample image respectively comprises the following steps:
filtering the image using a least squares filter;
dividing the filtered image into a plurality of scales;
performing subtraction computation among different scales to obtain detail information with different degrees;
weighting detail information with different degrees into an original image;
and obtaining the enhanced image containing abundant detail information.
3. The K-means worker fatigue simple discrimination method based on the multi-feature operators according to claim 1, wherein the SIFT feature extraction and the SIFT feature operator are used for achieving image characterization, the GIST feature extraction and the GIST feature operator are used for achieving image characterization and HOG feature extraction, and the HOG feature operator is used for achieving image characterization parallel operation.
4. The K-means worker fatigue simple discrimination method based on the multi-feature operator according to claim 1, wherein SIFT feature extraction comprises:
constructing a scale space; the scale space of the defined image is:
；
wherein G is a gaussian function:
；
wherein G (x, y, sigma) is a scale variable Gaussian function, (x, y) is a space coordinate, sigma is a scale space factor, sigma is a standard deviation of Gaussian normal distribution, reflects the degree of blurring of an image, and the larger the value is, the more blurring of the image is, the larger the corresponding scale is, and L (x, y, sigma) corresponds to Gaussian scale space.
5. The method for simply distinguishing the fatigue of the K-means workers based on the multi-feature operator according to claim 1, wherein the step of putting the characterization image, the plurality of positive training sample images and the plurality of negative training sample images together is characterized in that the number of the positive training sample images is identical to the number of the negative training sample images.
6. The simple and easy discriminating method of K-means workers' fatigue based on multi-feature operators according to claim 1 wherein the number of positive training samples and negative training samples is at least 50 respectively.
7. The simple and easy discriminating method of K-means workers' fatigue based on multi-feature operators according to claim 1, wherein the number of positive training samples and negative training samples is 100 respectively.
8. The simple and easy discriminating method of K-means workers' fatigue based on multi-feature operators as defined in claim 1, wherein said selecting method comprises: and (5) manual selection.
9. The simple and easy discriminating method of K-means workers' fatigue based on multi-feature operators as defined in claim 1, wherein said selecting method comprises: and (5) selecting a machine.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110686803.7A CN113344095B (en) | 2021-06-21 | 2021-06-21 | K-means worker fatigue simple discrimination method based on multi-feature operator |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110686803.7A CN113344095B (en) | 2021-06-21 | 2021-06-21 | K-means worker fatigue simple discrimination method based on multi-feature operator |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113344095A CN113344095A (en) | 2021-09-03 |
| CN113344095B true CN113344095B (en) | 2024-03-29 |
Family
ID=77478347
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110686803.7A Active CN113344095B (en) | 2021-06-21 | 2021-06-21 | K-means worker fatigue simple discrimination method based on multi-feature operator |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113344095B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105138982A (en) * | 2015-08-21 | 2015-12-09 | 中南大学 | Crowd abnormity detection and evaluation method based on multi-characteristic cluster and classification |
| CN111291678A (en) * | 2020-02-06 | 2020-06-16 | 北京爱笔科技有限公司 | Face image clustering method and device based on multi-feature fusion |
| CN112528815A (en) * | 2020-12-05 | 2021-03-19 | 西安电子科技大学 | Fatigue driving detection method based on multi-mode information fusion |
| CN112617772A (en) * | 2021-01-05 | 2021-04-09 | 上海工程技术大学 | Driving fatigue identification method and system based on pulse wave signals |
| WO2021068781A1 (en) * | 2019-10-12 | 2021-04-15 | 广东电网有限责任公司电力科学研究院 | Fatigue state identification method, apparatus and device |
-
2021
- 2021-06-21 CN CN202110686803.7A patent/CN113344095B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105138982A (en) * | 2015-08-21 | 2015-12-09 | 中南大学 | Crowd abnormity detection and evaluation method based on multi-characteristic cluster and classification |
| WO2021068781A1 (en) * | 2019-10-12 | 2021-04-15 | 广东电网有限责任公司电力科学研究院 | Fatigue state identification method, apparatus and device |
| CN111291678A (en) * | 2020-02-06 | 2020-06-16 | 北京爱笔科技有限公司 | Face image clustering method and device based on multi-feature fusion |
| CN112528815A (en) * | 2020-12-05 | 2021-03-19 | 西安电子科技大学 | Fatigue driving detection method based on multi-mode information fusion |
| CN112617772A (en) * | 2021-01-05 | 2021-04-09 | 上海工程技术大学 | Driving fatigue identification method and system based on pulse wave signals |
Non-Patent Citations (2)
| Title |
|---|
| 机动车驾驶员疲劳状态识别***研究;刘海平;《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》(第3期);全文 * |
| 车载疲劳驾驶行为检测***的研究与设计;聂尔豪;苏维均;于重重;郑雅;;计算机仿真(08);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113344095A (en) | 2021-09-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114549522B (en) | Textile quality detection method based on target detection | |
| CN104408449B (en) | Intelligent mobile terminal scene literal processing method | |
| CN109918971B (en) | Method and device for detecting number of people in monitoring video | |
| CN104408429A (en) | Method and device for extracting representative frame of video | |
| CN109948625A (en) | Definition of text images appraisal procedure and system, computer readable storage medium | |
| CN107909081A (en) | The quick obtaining and quick calibrating method of image data set in a kind of deep learning | |
| CN114723704A (en) | Textile quality evaluation method based on image processing | |
| CN113205081B (en) | SVM model worker fatigue accurate judging method based on significance detection | |
| CN109948566A (en) | A kind of anti-fraud detection method of double-current face based on weight fusion and feature selecting | |
| Amelio et al. | A genetic algorithm for color image segmentation | |
| Galsgaard et al. | Circular hough transform and local circularity measure for weight estimation of a graph-cut based wood stack measurement | |
| CN107886066A (en) | A kind of pedestrian detection method based on improvement HOG SSLBP | |
| Wen et al. | Virus image classification using multi-scale completed local binary pattern features extracted from filtered images by multi-scale principal component analysis | |
| CN108830842A (en) | A kind of medical image processing method based on Corner Detection | |
| CN106971158A (en) | A kind of pedestrian detection method based on CoLBP symbiosis feature Yu GSS features | |
| CN111428701B (en) | Small-area fingerprint image feature extraction method, system, terminal and storage medium | |
| CN116452506A (en) | Underground gangue intelligent visual identification and separation method based on machine learning | |
| CN115272664A (en) | Instrument panel display method and device, electronic equipment and storage medium | |
| CN114241522A (en) | Method, system, equipment and storage medium for field operation safety wearing identification | |
| CN112381140B (en) | Abrasive particle image machine learning identification method based on new characteristic parameters | |
| CN113344095B (en) | K-means worker fatigue simple discrimination method based on multi-feature operator | |
| CN108549868A (en) | A kind of pedestrian detection method | |
| CN112734747A (en) | Target detection method and device, electronic equipment and storage medium | |
| CN103679170B (en) | Method for detecting salient regions based on local features | |
| Caraka et al. | Batik parang rusak detection using geometric invariant moment |
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 |