CN109101922A - Operating personnel device, assay, device and electronic equipment - Google Patents
Operating personnel device, assay, device and electronic equipment Download PDFInfo
- Publication number
- CN109101922A CN109101922A CN201810913117.7A CN201810913117A CN109101922A CN 109101922 A CN109101922 A CN 109101922A CN 201810913117 A CN201810913117 A CN 201810913117A CN 109101922 A CN109101922 A CN 109101922A
- Authority
- CN
- China
- Prior art keywords
- histogram
- cell
- color
- human body
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/50—Extraction of image or video features by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides a kind of operating personnel device, assay, device and electronic equipments, are related to image identification technical field, this method comprises: obtaining the human body image of RGB color;Human body image is divided into multiple cells according to predetermined structure proportion;The histogram of gradients of the cell is generated according to the pixel value of pixel each in each cell;The color histogram of the cell is generated in the color value of RGB color according to pixel each in each cell;Classifier trained according to the histogram of gradients of each unit lattice and color histogram and in advance determines the corresponding dressing analysis result of human body image.Human body image is resolved by multiple cells according to the difference of real human body and apparel construction in this way, local shape factor is carried out to each cell based on the color and shape of clothes again, greatly reduce computation complexity, the recognition accuracy for improving dressing analysis, enhances the robustness in complex environment.
Description
Technical field
The present invention relates to image identification technical field, more particularly, to a kind of operating personnel device, assay, device and
Electronic equipment.
Background technique
With video monitoring system networking, intelligentized rapid development, it is public that intelligent video monitoring system has become guarantee
Safety altogether realizes the important channel of real time monitoring analysis.Using image sequence as input, major function includes target for intelligent monitoring
The processes such as detection, identification and tracking, realize the behavioural analysis to target object, and carry out safety alarm as needed.
Currently, preferable implementation and application is had been obtained to the identification and behavioural analysis of human body based on video system,
And the dressing of operating personnel is analyzed in substation, on the one hand because it is for the sensibility of complex environment and noise and higher
Calculating cost, do not developed completely also in practical applications;On the other hand, often occur blocking under power transformation operation environment,
The case where impact analysis such as uneven illumination, size and visual angle change detect greatly reduces the recognition accuracy of dressing analysis.
In summary, existing operating personnel device, assay also need further to improve, recognition accuracy need
It further increases.
Summary of the invention
In view of this, the purpose of the present invention is to provide a kind of operating personnel device, assay, device and electronic equipment,
To improve the recognition accuracy of dressing analysis, enhance the robustness in complex environment.
In a first aspect, the embodiment of the invention provides a kind of operating personnel device, assays, comprising:
The human body image of RGB color is obtained, the human body image is in monitoring video frame to be analyzed through background subtraction
Image afterwards;The human body image is divided into multiple cells according to predetermined structure proportion;According to each unit
The pixel value of each pixel generates the histogram of gradients of the cell in lattice;According to each pixel in each cell
Point generates the color histogram of the cell in the color value of the RGB color;According to the ladder of each cell
Histogram and color histogram and classifier trained in advance are spent, determines the corresponding dressing analysis result of the human body image;
Wherein, the classifier is that histogram of gradients and color histogram training based on training sample obtain.
With reference to first aspect, the embodiment of the invention provides the first possible embodiments of first aspect, wherein institute
It states and the human body image is divided into multiple cells according to predetermined structure proportion, comprising:
The human body image is divided into according to predetermined structure proportion to three for respectively corresponding the helmet, jacket and lower clothing
Cell.
With reference to first aspect, the embodiment of the invention provides second of possible embodiments of first aspect, wherein institute
State the histogram of gradients that the cell is generated according to the pixel value of each pixel in each cell, comprising:
According to the pixel value of each pixel in each cell, the gradient width of each pixel is calculated
Value and gradient direction;Corresponding gradient magnitude is mapped to according to the gradient direction of each pixel in each cell more
The first passage of the different gradient direction ranges of a preset correspondence, generates the histogram of gradients of the cell.
With reference to first aspect, the embodiment of the invention provides the third possible embodiments of first aspect, wherein institute
State the color for generating the cell in the color value of the RGB color according to each pixel in each cell
Histogram, comprising:
Each pixel in each cell is transformed into hsv color sky in the color value of the RGB color
Between in, obtain the tone value and intensity value of each pixel in the cell;By each pixel in each cell
The tone value of point is mapped to the second channel of multiple preset correspondences different tone ranges and saturation degree range with intensity value, raw
At the color histogram of the cell.
With reference to first aspect, the embodiment of the invention provides the 4th kind of possible embodiments of first aspect, wherein institute
The histogram of gradients and color histogram according to each cell and classifier trained in advance are stated, determines the people
Result is analyzed in the corresponding dressing of body image, comprising:
The histogram of gradients of each cell and color histogram are got up according to setting sequential series, obtained described
The target feature vector of human body image;By in target feature vector input classifier trained in advance, the classification is obtained
Result is analyzed in the corresponding dressing of the human body image of device output.
With reference to first aspect, the embodiment of the invention provides the 5th kind of possible embodiments of first aspect, wherein institute
The method of stating further includes by the following procedure training classifier:
Training sample is obtained, the training sample includes multiple correct dressing samples and multiple incorrect dressing samples;It is raw
At the histogram of gradients and color histogram of the training sample;According to the histogram of gradients and color histogram of the training sample
Figure determines the training feature vector of the training sample;Two disaggregated models are trained using the training feature vector, with
Obtain the classifier.
Second aspect, the embodiment of the present invention also provide a kind of operating personnel's dressing analytical equipment, comprising:
Image collection module, for obtaining the human body image of RGB color, the human body image is monitoring to be analyzed view
Image in frequency frame after background subtraction;Picture breakdown module is used for the human body image according to predetermined structure ratio
Example is divided into multiple cells;First generation module, for being generated according to the pixel value of each pixel in each cell
The histogram of gradients of the cell;Second generation module is used for according to each pixel in each cell described
The color value of RGB color generates the color histogram of the cell;As a result determining module, for according to each list
The histogram of gradients and color histogram of first lattice and classifier trained in advance, determine the corresponding dressing of the human body image
Analyze result;Wherein, the classifier is that histogram of gradients and color histogram training based on training sample obtain.
In conjunction with second aspect, the embodiment of the invention provides the first possible embodiments of second aspect, wherein institute
Picture breakdown module is stated to be specifically used for:
The human body image is divided into according to predetermined structure proportion to three for respectively corresponding the helmet, jacket and lower clothing
Cell.
In conjunction with second aspect, the embodiment of the invention provides the first possible embodiments of second aspect, wherein institute
Stating device further includes training module, for passing through the following procedure training classifier:
Training sample is obtained, the training sample includes multiple correct dressing samples and multiple incorrect dressing samples;It is raw
At the histogram of gradients and color histogram of the training sample;According to the histogram of gradients and color histogram of the training sample
Figure determines the training feature vector of the training sample;Two disaggregated models are trained using the training feature vector, with
Obtain the classifier.
The third aspect, the embodiment of the present invention also provide a kind of electronic equipment, including memory, processor, the memory
In be stored with the computer program that can be run on the processor, the processor is realized when executing the computer program
State method described in first aspect or its any possible embodiment.
The embodiment of the present invention bring it is following the utility model has the advantages that
In the embodiment of the present invention, the human body image of RGB color is obtained, human body image is monitoring video frame to be analyzed
The middle image after background subtraction;Human body image is divided into multiple cells according to predetermined structure proportion;According to every
The pixel value of each pixel generates the histogram of gradients of the cell in a cell;According to pixel each in each cell
Point generates the color histogram of the cell in the color value of RGB color;According to the histogram of gradients of each unit lattice and
Color histogram and classifier trained in advance determine the corresponding dressing analysis result of human body image;Wherein, the classification
Device is that the histogram of gradients and color histogram training based on training sample obtain.In this way according to real human body and apparel construction
Difference human body image is resolved into multiple cells, then the color and shape based on clothes respectively to each cell carry out office
Portion's feature extraction, greatly reduces computation complexity, improves the recognition accuracy of dressing analysis, enhances in complex environment
Robustness.
Other features and advantages of the present invention will illustrate in the following description, also, partly become from specification
It obtains it is clear that understand through the implementation of the invention.The objectives and other advantages of the invention are in specification and attached drawing
Specifically noted structure is achieved and obtained.
To enable the above objects, features and advantages of the present invention to be clearer and more comprehensible, preferred embodiment is cited below particularly, and cooperate
Appended attached drawing, is described in detail below.
Detailed description of the invention
It, below will be to specific in order to illustrate more clearly of the specific embodiment of the invention or technical solution in the prior art
Embodiment or attached drawing needed to be used in the description of the prior art be briefly described, it should be apparent that, it is described below
Attached drawing is some embodiments of the present invention, for those of ordinary skill in the art, before not making the creative labor
It puts, is also possible to obtain other drawings based on these drawings.
Fig. 1 be a kind of operating personnel provided in an embodiment of the present invention the flow diagram of device, assay;
Fig. 2 is a kind of flow diagram of trained classifier provided in an embodiment of the present invention;
Fig. 3 is a kind of structural schematic diagram of operating personnel's dressing analytical equipment provided in an embodiment of the present invention;
Fig. 4 is the structural schematic diagram of a kind of electronic equipment provided in an embodiment of the present invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with attached drawing to the present invention
Technical solution be clearly and completely described, it is clear that described embodiments are some of the embodiments of the present invention, rather than
Whole embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art are not making creative work premise
Under every other embodiment obtained, shall fall within the protection scope of the present invention.
Current existing operating personnel device, assay also need further to improve, recognition accuracy needs further
It improves.Based on this, a kind of operating personnel based on S-HOG+C operator provided in an embodiment of the present invention device, assay, device
And electronic equipment, the recognition accuracy of dressing analysis can be improved, enhance the robustness in complex environment.
To divide a kind of operating personnel dressing disclosed in the embodiment of the present invention first convenient for understanding the present embodiment
Analysis method describes in detail.
Embodiment one:
Fig. 1 be a kind of operating personnel provided in an embodiment of the present invention the flow diagram of device, assay, this method adopt
With S-HOG+C operator, wherein S (Structure), which is represented, carries out the division of human body image structure, HOG (Histogram of
Oriented Gradients) indicate gradient orientation histogram (may be simply referred to as histogram of gradients), HOC (Histogram of
Color color histogram) is indicated.As shown in Figure 1, this method including the following steps:
Step S101, obtains the human body image of RGB color, and human body image is in monitoring video frame to be analyzed through carrying on the back
Image after scape subduction.
Above-mentioned human body image can be the video frame randomly selected in the video sequence of monitor video after background subtraction
Image, wherein in the video frame carry out background subtraction process be referred to related art, it is not limited here
It is fixed.
Above-mentioned human body image is divided into multiple cells according to predetermined structure proportion by step S102.
Above structure ratio refers to the structure proportion of human body and clothes, which can be arranged according to the actual situation.
In view of the shape feature of the helmet of operating personnel (such as power transformation operation personnel), jacket and lower three parts of clothing is obvious
Difference, and differ greatly between each section, in some possible embodiments, detailed process is as follows by step S102: will be above-mentioned
Human body image is divided into three cells for respectively corresponding the helmet, jacket and lower clothing according to predetermined structure proportion, remembers respectively
For (C1, C2, C3)。
Specifically, the structure proportion of predetermined human body and clothes can be 1:4:4, and human body image is divided into helmet list
First lattice, jacket cell and lower clothing cell.
Step S103 generates the histogram of gradients of the cell according to the pixel value of pixel each in each cell.
In some possible embodiments, above-mentioned steps S103 specifically: according to pixel each in each cell
The gradient magnitude and gradient direction of each pixel is calculated in pixel value;According to the ladder of pixel each in each cell
Corresponding gradient magnitude is mapped to the first passage of the different gradient direction ranges of multiple preset correspondences, generation unit by degree direction
The histogram of gradients of lattice.
Optionally, above-mentioned multiple first passages can be 9 channels for dividing 0 ° to 180 ° of gradient direction range equally, often
A channel span is 20 °.
Step S104, the color value according to pixel each in each cell in RGB color generate the cell
Color histogram.
In some possible embodiments, above-mentioned steps S104 specifically: by pixel each in each cell in RGB
The color value of color space transforms in HSV (Hue, Saturation, Value) color space, obtains each in the cell
The tone value and intensity value of pixel;By the tone value of pixel each in each cell and intensity value (tone and full
With the combination of degree) it is mapped to the second channel of multiple preset correspondences different tone ranges and saturation degree range, generate the unit
The color histogram of lattice.
Optionally, tone range (0 ° to 360 °) and saturation degree range (0%~100%) are evenly dividing respectively is 5
Channel, combination of two share 5 × 5=25 kind combination of channels, i.e., multiple second channels are 25 channels of corresponding 25 kinds of combinations.
It should be noted that executing sequence without successive between above-mentioned steps S103 and S104.
Step S105, classifier trained according to the histogram of gradients of each unit lattice and color histogram and in advance,
Determine the corresponding dressing analysis result of above-mentioned human body image.
Wherein, above-mentioned classifier is that histogram of gradients and color histogram training based on training sample obtain.
Specifically, the histogram of gradients of each unit lattice and color histogram are got up according to setting sequential series, is obtained
The target feature vector of above-mentioned human body image;By in target feature vector input classifier trained in advance, the classification is obtained
Result is analyzed in the corresponding dressing of human body image of device output.
With cell Ci(i=1,2,3), gradient magnitude is mapped to for 9 channels, generation unit lattice CiGradient it is straight
The process of square figure can be with are as follows:
(G1) each unit lattice, that is, C is calculatediThe gradient magnitude of each pixel in (i=1,2,3)With gradient direction Θ.
Such as:
(G1-1) gradient template of the discrete point of an One-Dimensional Center is applied in the horizontal and vertical directions respectively, it can be with
Convolution is carried out to the corresponding level matrix of the current pixel point traversed and vertical matrix respectively using following formula (1) convolution kernel, from
And obtain the horizontal component v of current pixel pointxWith vertical component vy;Wherein, level matrix is by current pixel point horizontal direction
The pixel value of two neighbor pixels and the pixel value of current pixel point are constituted, and vertical matrix is by current pixel point vertical direction
Two neighbor pixels pixel value and current pixel point pixel value constitute.
[- 1,0,1] and [- 1,0,1]T (1)
(G1-2) whole image window (each pixel of cell) is traversed, calculates the gradient magnitude of pixel pIt is shown with gradient direction Θ (p) such as following formula (2) and formula (3):
Wherein, Θ (p) is the unsigned real that codomain is 0 ° to 180 °.
(G2) after corresponding gradient magnitude being mapped to 9 channels according to the gradient direction of each pixel, normalizing is generated
S-HOG histogram hog after changei(i=1,2,3).
Specifically, by each unit lattice Ci(i=1,2,3) is divided into the less multiple homalographic cell block (lists of pixel
First block can be rectangle), to pixel each in cell block according to its gradient direction by its amplitudeIt is accumulated in corresponding logical
On road, and then the histogram for containing 9 bin is constructed to each cell block;Multiple (such as 2*2) cell blocks are constituted bigger
Extent block, obtain the histogram of extent block, and independent feature vector normalization is carried out to the histogram of extent block, returned
Histogram after one change;Finally obtained according to the histogram after all extent blocks normalization of setting sequence (can manually be arranged) series connection
To S-HOG histogram, i.e. histogram of gradients hogi(i=1,2,3).
The size of said units block can be arranged according to actual needs, and the corresponding cell block of different units lattice can be set
It is identical at area, area difference also can be set into.Such as it can be by each unit lattice CiThe corresponding cell block face (i=1,2,3)
Product is disposed as 10*10, can also be by C1Corresponding cell block area is set as 10*10, C2、C3Corresponding cell block area is set
It is set to 20*20.
With cell Ci(i=1,2,3), by for tone and the combinatorial mapping of saturation degree to 25 channels, generation unit
Lattice CiThe process of color histogram can be with are as follows:
(C1) by CiTransform to hsv color space hsiIn, one by one by hsiIn pixel respectively according to tone value and saturation
Angle value is uniformly assigned in 5 channels, and the tone of each pixel and the combination of saturation degree are obtained.
(C2) by hsiIn after the tone of each pixel and the combinatorial mapping to 25 channels of saturation degree, generate normalization
S-HOC histogram hoc afterwardsi(i=1,2,3).
Specifically the S-HOC histogram after normalizing can be generated according to the method for generating normalization S-HOG histogram, this
In repeat no more.
Based on above-mentioned example, determine that the process of the corresponding dressing analysis result of above-mentioned human body image can be with are as follows:
The feature vector S-HOG+C (target feature vector) that obtain six histograms are together in series to the end, tool
Body, feature vector S-HOG+C=[hog1, hog2, hog3, hoc1, hoc2, hoc3], wherein hogi(i=1,2,3) is i-th
The histogram of gradients of cell, hoci(i=1,2,3) be i-th of cell color histogram, share the helmet, jacket and under
Three cells of clothing.The S-HOG+C feature vector is then based on using classifier trained in advance to differentiate above-mentioned human body image
Whether corresponding operating personnel's dressing meets dress code, obtains dressing analysis result.
To sum up, in the embodiment of the present invention, the human body image of RGB color is obtained, human body image is monitoring to be analyzed
Image in video frame after background subtraction;Human body image is divided into multiple cells according to predetermined structure proportion;
The histogram of gradients of the cell is generated according to the pixel value of pixel each in each cell;According to each in each cell
A pixel generates the color histogram of the cell in the color value of RGB color;Gradient according to each unit lattice is straight
Side's figure and color histogram and classifier trained in advance determine the corresponding dressing analysis result of human body image;Wherein,
The classifier is that the histogram of gradients and color histogram training based on training sample obtain.In this way according to real human body kimonos
Human body image is resolved into multiple cells by the difference of assembling structure, then is carried out based on the color and shape of clothes to each cell
Local shape factor greatly reduces computation complexity, improves the recognition accuracy of dressing analysis, enhances in complex environment
In robustness.
Fig. 2 is a kind of flow diagram of trained classifier provided in an embodiment of the present invention, as shown in Fig. 2, by following
Process trains classifier:
Step S201 obtains training sample, which includes multiple correct dressing samples and multiple incorrect dressings
Sample.
In order to make classifier that there is preferably classifying quality, need using enough correct dressing sample (positive class) and non-
Correct dressing sample (negative class) goes to train classifier.
Step S202 generates the histogram of gradients and color histogram of above-mentioned training sample.
Referring to the process for generating histogram of gradients and color histogram in above-mentioned Fig. 1, each training sample is generated respectively
Histogram of gradients and color histogram.
Step S203 determines the training of the training sample according to the histogram of gradients of above-mentioned training sample and color histogram
Feature vector.
Referring to the process of target feature vector is obtained in above-mentioned Fig. 1, the training feature vector of training sample is determined.
Step S204 is trained two disaggregated models using above-mentioned training feature vector, to obtain classifier.
Two disaggregated models can be, but not limited to support vector machines (SVM, support vector machines).
In order to verify operating personnel provided in this embodiment the effect of device, assay, with the operation people in certain substation
Experiment test has been carried out for member.It will wearing yellow safety cap, blue working coat and the lower clothing definition of blue work in the test
For correct dressing sample, that is, be positive class, will wear other color-safe caps or not have safe wearing cap, jacket or lower clothing not
The definition for being blue work clothes is incorrect dressing, that is, be negative class.All sample datas are selected in the case where blocking less scene
Shooting is extracted, and selects linear SVM to train classifier.
Identification in order to illustrate operating personnel provided in this embodiment device, assay to other positive classes and negative class sample
Effect is counted and is evaluated dressing analysis using three accuracy (ACC), true positive rate (TPR) and false positive rate (FPR) indexs
Recognition accuracy, the definition of these three indexs is as follows:
Wherein, P and N respectively indicates the number of samples of positive class and negative class in sample, and TP indicates prediction result and label is all
The quantity of positive class, TN indicate that prediction result and label are all negative the quantity of class, and FP indicates that the prediction category label that are positive are negative the sample of class
This quantity.
Using classifier recognition effect in this experiment of above three metrics evaluation, the results are shown in Table 1:
1 operating personnel's dressing of table analysis experiment accuracy rate (%)
In table 1, positive training condition indicates the case where carrying out discriminance analysis to the positive dressing image of operating personnel, laterally
Training condition indicates the case where carrying out discriminance analysis to the lateral dressing image of operating personnel, and positive side combined training condition indicates
The case where discriminance analysis is carried out to operating personnel's forward direction and lateral dressing image.As it can be seen from table 1 in positive unobstructed feelings
Under condition, the precision of experiment even can achieve 98.33%;The FPR under three kinds of different training conditions is 0 simultaneously, this table
The proposed method of the present embodiment is illustrated and is minimizing the outstanding behaviours in error.Table 1 demonstrates the method that the present embodiment is proposed
Validity, and its stability to staff's direction variation.
Embodiment two:
Fig. 3 is a kind of structural schematic diagram of operating personnel's dressing analytical equipment provided in an embodiment of the present invention, such as Fig. 3 institute
Show, which includes:
Image collection module 31, for obtaining the human body image of RGB color, human body image is monitoring to be analyzed view
Image in frequency frame after background subtraction;
Picture breakdown module 32, for above-mentioned human body image to be divided into multiple units according to predetermined structure proportion
Lattice;
First generation module 33, for generating the ladder of the cell according to the pixel value of pixel each in each cell
Spend histogram;
Second generation module 34, for raw in the color value of RGB color according to pixel each in each cell
At the color histogram of the cell;
As a result determining module 35, for the histogram of gradients and color histogram and preparatory instruction according to each unit lattice
Experienced classifier determines the corresponding dressing analysis result of above-mentioned human body image;Wherein, which is the ladder based on training sample
What degree histogram and color histogram training obtained.
Optionally, above-mentioned picture breakdown module 32 is specifically used for:
Above-mentioned human body image is divided into according to predetermined structure proportion to three for respectively corresponding the helmet, jacket and lower clothing
Cell.
Optionally, above-mentioned first generation module 33 is specifically used for:
According to the pixel value of pixel each in each cell, the gradient magnitude and gradient of each pixel is calculated
Direction;Corresponding gradient magnitude is mapped to multiple preset correspondences according to the gradient direction of pixel each in each cell
The first passage of different gradient direction ranges, generates the histogram of gradients of the cell.
Optionally, above-mentioned second generation module 34 is specifically used for:
Color value by pixel each in each cell in RGB color transforms in hsv color space, obtains
The tone value and intensity value of each pixel in the cell;By the tone value and saturation of pixel each in each cell
Angle value is mapped to the second channel of multiple preset correspondences different tone ranges and saturation degree range, generates the color of the cell
Histogram.
Optionally, as a result determining module 35 is specifically used for:
The histogram of gradients of each unit lattice and color histogram are got up according to setting sequential series, obtain above-mentioned human body
The target feature vector of image;By in target feature vector input classifier trained in advance, classifier output is obtained
Result is analyzed in the corresponding dressing of human body image.
Above-mentioned apparatus further includes training module, for passing through the above-mentioned classifier of following procedure training:
Training sample is obtained, which includes multiple correct dressing samples and multiple incorrect dressing samples;It generates
The histogram of gradients and color histogram of the training sample;It is determined according to the histogram of gradients of the training sample and color histogram
The training feature vector of the training sample;Two disaggregated models are trained using the training feature vector, to obtain above-mentioned point
Class device.
In the embodiment of the present invention, the human body image of RGB color is obtained, human body image is monitoring video frame to be analyzed
The middle image after background subtraction;Human body image is divided into multiple cells according to predetermined structure proportion;According to every
The pixel value of each pixel generates the histogram of gradients of the cell in a cell;According to pixel each in each cell
Point generates the color histogram of the cell in the color value of RGB color;According to the histogram of gradients of each unit lattice and
Color histogram and classifier trained in advance determine the corresponding dressing analysis result of human body image;Wherein, the classification
Device is that the histogram of gradients and color histogram training based on training sample obtain.In this way according to real human body and apparel construction
Difference human body image is resolved into multiple cells, then based on the color and shape of clothes to carry out part to each cell special
Sign is extracted, and computation complexity is greatly reduced, and is improved the recognition accuracy of dressing analysis, is enhanced the Shandong in complex environment
Stick.
Embodiment three:
Referring to fig. 4, the embodiment of the present invention also provides a kind of electronic equipment 100, comprising: processor 40, memory 41, bus
42 and communication interface 43, the processor 40, communication interface 43 and memory 41 are connected by bus 42;Processor 40 is for holding
The executable module stored in line storage 41, such as computer program.
Wherein, memory 41 may include high-speed random access memory (RAM, RandomAccessMemory), can also
It can further include nonvolatile memory (non-volatile memory), for example, at least a magnetic disk storage.Pass through at least one
A communication interface 43 (can be wired or wireless) realizes the communication link between the system network element and at least one other network element
It connects, internet, wide area network, local network, Metropolitan Area Network (MAN) etc. can be used.
Bus 42 can be isa bus, pci bus or eisa bus etc..The bus can be divided into address bus, data
Bus, control bus etc..Only to be indicated with a four-headed arrow convenient for indicating, in Fig. 4, it is not intended that an only bus or
A type of bus.
Wherein, memory 41 is for storing program, and the processor 40 executes the journey after receiving and executing instruction
Sequence, method performed by the device that the stream process that aforementioned any embodiment of the embodiment of the present invention discloses defines can be applied to handle
In device 40, or realized by processor 40.
Processor 40 may be a kind of IC chip, the processing capacity with signal.During realization, above-mentioned side
Each step of method can be completed by the integrated logic circuit of the hardware in processor 40 or the instruction of software form.Above-mentioned
Processor 40 can be general processor, including central processing unit (Central Processing Unit, abbreviation CPU), network
Processor (Network Processor, abbreviation NP) etc.;It can also be digital signal processor (Digital Signal
Processing, abbreviation DSP), specific integrated circuit (Application Specific Integrated Circuit, referred to as
ASIC), ready-made programmable gate array (Field-Programmable Gate Array, abbreviation FPGA) or other are programmable
Logical device, discrete gate or transistor logic, discrete hardware components.It may be implemented or execute in the embodiment of the present invention
Disclosed each method, step and logic diagram.General processor can be microprocessor or the processor is also possible to appoint
What conventional processor etc..The step of method in conjunction with disclosed in the embodiment of the present invention, can be embodied directly in hardware decoding processing
Device executes completion, or in decoding processor hardware and software module combination execute completion.Software module can be located at
Machine memory, flash memory, read-only memory, programmable read only memory or electrically erasable programmable memory, register etc. are originally
In the storage medium of field maturation.The storage medium is located at memory 41, and processor 40 reads the information in memory 41, in conjunction with
Its hardware completes the step of above method.
Operating personnel's dressing analytical equipment and electronic equipment provided in an embodiment of the present invention, with work provided by the above embodiment
Industry personnel device, assay technical characteristic having the same reach identical skill so also can solve identical technical problem
Art effect.
It is apparent to those skilled in the art that for convenience and simplicity of description, the device of foregoing description
And the specific work process of electronic equipment, it can refer to corresponding processes in the foregoing method embodiment, details are not described herein.
Unless specifically stated otherwise, the opposite step of the component and step that otherwise illustrate in these embodiments, digital table
It is not limit the scope of the invention up to formula and numerical value.
In all examples being illustrated and described herein, any occurrence should be construed as merely illustratively, without
It is as limitation, therefore, other examples of exemplary embodiment can have different values.
The flow chart and block diagram in the drawings show the device of multiple embodiments according to the present invention, method and computer journeys
The architecture, function and operation in the cards of sequence product.In this regard, each box in flowchart or block diagram can generation
A part of one module, section or code of table, a part of the module, section or code include one or more use
The executable instruction of the logic function as defined in realizing.It should also be noted that in some implementations as replacements, being marked in box
The function of note can also occur in a different order than that indicated in the drawings.For example, two continuous boxes can actually base
Originally it is performed in parallel, they can also be executed in the opposite order sometimes, and this depends on the function involved.It is also noted that
It is the combination of each box in block diagram and or flow chart and the box in block diagram and or flow chart, can uses and execute rule
The dedicated hardware based system of fixed function or movement is realized, or can use the group of specialized hardware and computer instruction
It closes to realize.
Carry out provided by the embodiment of the present invention operating personnel the computer program product of device, assay, including storage
The computer readable storage medium of the executable non-volatile program code of processor, the instruction that said program code includes can
For executing previous methods method as described in the examples, specific implementation can be found in embodiment of the method, and details are not described herein.
In several embodiments provided herein, it should be understood that disclosed device and method can pass through it
Its mode is realized.The apparatus embodiments described above are merely exemplary, for example, the division of the unit, only
A kind of logical function partition, there may be another division manner in actual implementation, in another example, multiple units or components can combine
Or it is desirably integrated into another system, or some features can be ignored or not executed.Another point, shown or discussed phase
Coupling, direct-coupling or communication connection between mutually can be through some communication interfaces, the INDIRECT COUPLING of device or unit or
Communication connection can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.
It, can be with if the function is realized in the form of SFU software functional unit and when sold or used as an independent product
It is stored in the executable non-volatile computer-readable storage medium of a processor.Based on this understanding, of the invention
Technical solution substantially the part of the part that contributes to existing technology or the technical solution can be with software in other words
The form of product embodies, which is stored in a storage medium, including some instructions use so that
One computer equipment (can be personal computer, server or the network equipment etc.) executes each embodiment institute of the present invention
State all or part of the steps of method.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (ROM, Read-
Only Memory), random access memory (RAM, Random Access Memory), magnetic or disk etc. are various can be with
Store the medium of program code.
Finally, it should be noted that embodiment described above, only a specific embodiment of the invention, to illustrate the present invention
Technical solution, rather than its limitations, scope of protection of the present invention is not limited thereto, although with reference to the foregoing embodiments to this hair
It is bright to be described in detail, those skilled in the art should understand that: anyone skilled in the art
In the technical scope disclosed by the present invention, it can still modify to technical solution documented by previous embodiment or can be light
It is readily conceivable that variation or equivalent replacement of some of the technical features;And these modifications, variation or replacement, do not make
The essence of corresponding technical solution is detached from the spirit and scope of technical solution of the embodiment of the present invention, should all cover in protection of the invention
Within the scope of.Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (10)
1. a kind of operating personnel device, assay characterized by comprising
Obtain RGB color human body image, the human body image be monitoring video frame to be analyzed in after background subtraction
Image;
The human body image is divided into multiple cells according to predetermined structure proportion;
The histogram of gradients of the cell is generated according to the pixel value of each pixel in each cell;
The face of the cell is generated in the color value of the RGB color according to each pixel in each cell
Color Histogram;
According to the histogram of gradients of each cell and color histogram and in advance trained classifier, determine described in
Result is analyzed in the corresponding dressing of human body image;Wherein, the classifier is that histogram of gradients based on training sample and color are straight
What side's figure training obtained.
2. the method according to claim 1, wherein it is described by the human body image according to predetermined structure
Ratio is divided into multiple cells, comprising:
The human body image is divided into three units for respectively corresponding the helmet, jacket and lower clothing according to predetermined structure proportion
Lattice.
3. the method according to claim 1, wherein described according to each pixel in each cell
Pixel value generates the histogram of gradients of the cell, comprising:
According to the pixel value of each pixel in each cell, be calculated each pixel gradient magnitude and
Gradient direction;
Corresponding gradient magnitude is mapped to according to the gradient direction of each pixel in each cell multiple preset
The first passage of corresponding different gradient direction ranges, generates the histogram of gradients of the cell.
4. the method according to claim 1, wherein described exist according to each pixel in each cell
The color value of the RGB color generates the color histogram of the cell, comprising:
Each pixel in each cell is transformed in hsv color space in the color value of the RGB color,
Obtain the tone value and intensity value of each pixel in the cell;
It is not homochromy that the tone value of each pixel and intensity value in each cell are mapped to multiple preset correspondences
The second channel for adjusting range and saturation degree range, generates the color histogram of the cell.
5. the method according to claim 1, wherein the histogram of gradients according to each cell and
Color histogram and classifier trained in advance determine the corresponding dressing analysis result of the human body image, comprising:
The histogram of gradients of each cell and color histogram are got up according to setting sequential series, obtain the human body
The target feature vector of image;
By in target feature vector input classifier trained in advance, the human body image of the classifier output is obtained
Result is analyzed in corresponding dressing.
6. the method according to claim 1, wherein the method also includes passing through described point of following procedure training
Class device:
Training sample is obtained, the training sample includes multiple correct dressing samples and multiple incorrect dressing samples;
Generate the histogram of gradients and color histogram of the training sample;
The training feature vector of the training sample is determined according to the histogram of gradients of the training sample and color histogram;
Two disaggregated models are trained using the training feature vector, to obtain the classifier.
7. a kind of operating personnel's dressing analytical equipment characterized by comprising
Image collection module, for obtaining the human body image of RGB color, the human body image is monitoring video frame to be analyzed
The middle image after background subtraction;
Picture breakdown module, for the human body image to be divided into multiple cells according to predetermined structure proportion;
First generation module, for generating the ladder of the cell according to the pixel value of each pixel in each cell
Spend histogram;
Second generation module, for according to each pixel in each cell in the color value of the RGB color
Generate the color histogram of the cell;
As a result determining module, for the histogram of gradients and color histogram and training in advance according to each cell
Classifier, determine the human body image corresponding dressing analysis result;Wherein, the classifier is the ladder based on training sample
What degree histogram and color histogram training obtained.
8. device according to claim 7, which is characterized in that described image decomposing module is specifically used for:
The human body image is divided into three units for respectively corresponding the helmet, jacket and lower clothing according to predetermined structure proportion
Lattice.
9. device according to claim 7, which is characterized in that described device further includes training module, for by following
The process training classifier:
Training sample is obtained, the training sample includes multiple correct dressing samples and multiple incorrect dressing samples;
Generate the histogram of gradients and color histogram of the training sample;
The training feature vector of the training sample is determined according to the histogram of gradients of the training sample and color histogram;
Two disaggregated models are trained using the training feature vector, to obtain the classifier.
10. a kind of electronic equipment, including memory, processor, it is stored with and can runs on the processor in the memory
Computer program, which is characterized in that the processor realizes any one of claim 1-6 when executing the computer program
The method.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810913117.7A CN109101922A (en) | 2018-08-10 | 2018-08-10 | Operating personnel device, assay, device and electronic equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810913117.7A CN109101922A (en) | 2018-08-10 | 2018-08-10 | Operating personnel device, assay, device and electronic equipment |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109101922A true CN109101922A (en) | 2018-12-28 |
Family
ID=64849515
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810913117.7A Pending CN109101922A (en) | 2018-08-10 | 2018-08-10 | Operating personnel device, assay, device and electronic equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109101922A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110427808A (en) * | 2019-06-21 | 2019-11-08 | 武汉倍特威视系统有限公司 | Police uniform recognition methods based on video stream data |
CN112257620A (en) * | 2020-10-27 | 2021-01-22 | 广州华微明天软件技术有限公司 | Safe wearing condition identification method |
CN113837138A (en) * | 2021-09-30 | 2021-12-24 | 重庆紫光华山智安科技有限公司 | Dressing monitoring method, system, medium and electronic terminal |
CN113920329A (en) * | 2021-11-08 | 2022-01-11 | 北京九州安华信息安全技术有限公司 | Feature extraction method and device based on gradient histogram |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103970771A (en) * | 2013-01-29 | 2014-08-06 | 中国科学院计算技术研究所 | Search method and system for human body |
CN103996046A (en) * | 2014-06-11 | 2014-08-20 | 北京邮电大学 | Personnel recognition method based on multi-visual-feature fusion |
CN104077605A (en) * | 2014-07-18 | 2014-10-01 | 北京航空航天大学 | Pedestrian search and recognition method based on color topological structure |
CN105160317A (en) * | 2015-08-31 | 2015-12-16 | 电子科技大学 | Pedestrian gender identification method based on regional blocks |
CN105224956A (en) * | 2015-10-16 | 2016-01-06 | 深圳市捷顺科技实业股份有限公司 | A kind of car plate classification based on multi-feature fusion recognition methods and system |
CN105678309A (en) * | 2016-02-03 | 2016-06-15 | 山东工商学院 | Image multi-tag marking algorithm based on multi-example package feature learning |
-
2018
- 2018-08-10 CN CN201810913117.7A patent/CN109101922A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103970771A (en) * | 2013-01-29 | 2014-08-06 | 中国科学院计算技术研究所 | Search method and system for human body |
CN103996046A (en) * | 2014-06-11 | 2014-08-20 | 北京邮电大学 | Personnel recognition method based on multi-visual-feature fusion |
CN104077605A (en) * | 2014-07-18 | 2014-10-01 | 北京航空航天大学 | Pedestrian search and recognition method based on color topological structure |
CN105160317A (en) * | 2015-08-31 | 2015-12-16 | 电子科技大学 | Pedestrian gender identification method based on regional blocks |
CN105224956A (en) * | 2015-10-16 | 2016-01-06 | 深圳市捷顺科技实业股份有限公司 | A kind of car plate classification based on multi-feature fusion recognition methods and system |
CN105678309A (en) * | 2016-02-03 | 2016-06-15 | 山东工商学院 | Image multi-tag marking algorithm based on multi-example package feature learning |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110427808A (en) * | 2019-06-21 | 2019-11-08 | 武汉倍特威视系统有限公司 | Police uniform recognition methods based on video stream data |
CN112257620A (en) * | 2020-10-27 | 2021-01-22 | 广州华微明天软件技术有限公司 | Safe wearing condition identification method |
CN112257620B (en) * | 2020-10-27 | 2021-10-26 | 广州华微明天软件技术有限公司 | Safe wearing condition identification method |
CN113837138A (en) * | 2021-09-30 | 2021-12-24 | 重庆紫光华山智安科技有限公司 | Dressing monitoring method, system, medium and electronic terminal |
CN113837138B (en) * | 2021-09-30 | 2023-08-29 | 重庆紫光华山智安科技有限公司 | Dressing monitoring method, dressing monitoring system, dressing monitoring medium and electronic terminal |
CN113920329A (en) * | 2021-11-08 | 2022-01-11 | 北京九州安华信息安全技术有限公司 | Feature extraction method and device based on gradient histogram |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109101922A (en) | Operating personnel device, assay, device and electronic equipment | |
CN110991465B (en) | Object identification method, device, computing equipment and storage medium | |
CN110070067A (en) | The training method of video classification methods and its model, device and electronic equipment | |
KR101640998B1 (en) | Image processing apparatus and image processing method | |
CN109447169A (en) | The training method of image processing method and its model, device and electronic system | |
EP3806064A1 (en) | Method and apparatus for detecting parking space usage condition, electronic device, and storage medium | |
CN105893925A (en) | Human hand detection method based on complexion and device | |
CN110443212B (en) | Positive sample acquisition method, device, equipment and storage medium for target detection | |
CN111027378A (en) | Pedestrian re-identification method, device, terminal and storage medium | |
CN108986075A (en) | A kind of judgment method and device of preferred image | |
CN105069774B (en) | The Target Segmentation method of optimization is cut based on multi-instance learning and figure | |
CN104036284A (en) | Adaboost algorithm based multi-scale pedestrian detection method | |
CN112613344B (en) | Vehicle track occupation detection method, device, computer equipment and readable storage medium | |
CN112819110B (en) | Incremental small sample target detection method and system based on weight generation | |
CN113989944B (en) | Operation action recognition method, device and storage medium | |
CN112613454A (en) | Electric power infrastructure construction site violation identification method and system | |
CN104854620A (en) | Image processing device, image processing system, and program | |
CN112287912A (en) | Deep learning-based lane line detection method and device | |
DK2447884T3 (en) | A method for the detection and recognition of an object in an image and an apparatus and a computer program therefor | |
CN106778731B (en) | A kind of license plate locating method and terminal | |
CN107330027A (en) | A kind of Weakly supervised depth station caption detection method | |
CN110826418A (en) | Face feature extraction method and device | |
CN109740609A (en) | A kind of gauge detection method and device | |
CN106651797A (en) | Determination method and apparatus for effective region of signal lamp | |
CN105631474A (en) | Hyperspectral data multi-class method based on Jeffries-Matusita distance and class pair decision tree |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181228 |
|
RJ01 | Rejection of invention patent application after publication |