CN109447942A - Image blur determines method, apparatus, computer equipment and storage medium - Google Patents
Image blur determines method, apparatus, computer equipment and storage medium Download PDFInfo
- Publication number
- CN109447942A CN109447942A CN201811073282.2A CN201811073282A CN109447942A CN 109447942 A CN109447942 A CN 109447942A CN 201811073282 A CN201811073282 A CN 201811073282A CN 109447942 A CN109447942 A CN 109447942A
- Authority
- CN
- China
- Prior art keywords
- pixel
- gradient
- image
- vertical
- target signature
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000003860 storage Methods 0.000 title claims abstract description 19
- 230000001815 facial effect Effects 0.000 claims abstract description 115
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 29
- 238000004590 computer program Methods 0.000 claims description 16
- 238000005520 cutting process Methods 0.000 claims description 8
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims 1
- 238000004364 calculation method Methods 0.000 description 11
- 230000006870 function Effects 0.000 description 11
- 238000010586 diagram Methods 0.000 description 10
- 238000006243 chemical reaction Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 230000010485 coping Effects 0.000 description 1
- 238000000151 deposition Methods 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011549 displacement method Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000005764 inhibitory process Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 235000015170 shellfish Nutrition 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- 238000013316 zoning Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/168—Feature extraction; Face representation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30196—Human being; Person
- G06T2207/30201—Face
Abstract
This application discloses a kind of image blurs to determine method, apparatus, computer equipment and storage medium, applied to field of image recognition, the judging efficiency of facial image fuzziness can be effectively improved, the application method part includes: to obtain target signature region grayscale image, and target signature region grayscale image is the corresponding grayscale image of target signature region of facial image;Horizontal direction convolution algorithm and vertical direction convolution algorithm are carried out to pixel each in target signature region grayscale image, to obtain the horizontal direction gradient and vertical gradient of each pixel;The corresponding approximate gradient of each pixel is determined according to the horizontal direction gradient and vertical gradient of each pixel;Objective fuzzy degree is determined according to the corresponding approximate gradient of each pixel, using objective fuzzy degree as the fuzziness of facial image.
Description
Technical field
This application involves field of image recognition more particularly to a kind of image blur to determine method, apparatus, computer equipment
And storage medium.
Background technique
It during recognition of face, needs to detect face, quality meets the people of subsequent processing requirement in order to obtain
Face image needs to need to carry out to the facial image after detection the process of Quality estimation, wherein judge its of quality of human face image
In one with reference to be exactly facial image fuzziness, currently used more is increase income computer vision library (Open Source
Computer Vision Library, opencv) Lai Jinhang facial image fuzziness judgement.
Wherein, opencv is a cross-platform computer vision library, may operate in the operations such as Linux, Windows system
On system.Its lightweight and efficiently --- be made of a series of C functions and a small amount of C++ class, while providing the language such as Python
Interface, realize many general-purpose algorithms in terms of image procossing and computer vision.But due to traditionally utilizing opencv
It is to be judged that calculation amount is larger based on whole face characteristic when the fuzziness of progress facial image judges, the judgement of fuzziness
Efficiency is relatively low.
Summary of the invention
Based on this, it is necessary to which in view of the above technical problems, facial image fuzziness can be effectively improved by providing one kind
The image blur of judging efficiency determines method, apparatus, computer equipment and storage medium.
A kind of image blur determines method, comprising:
Obtain facial image;
Facial image cut and gray proces are to obtain target signature region grayscale image, target signature region gray scale
Figure is the corresponding grayscale image of target signature region of facial image;
Horizontal direction convolution algorithm and vertical direction convolution fortune are carried out to pixel each in target signature region grayscale image
It calculates, to obtain the horizontal direction gradient and vertical gradient of each pixel;
The corresponding approximate ladder of each pixel is determined according to the horizontal direction gradient and vertical gradient of each pixel
Degree;
Objective fuzzy degree is determined according to the corresponding approximate gradient of each pixel, using objective fuzzy degree as facial image
Fuzziness.
A kind of image blur determining device, comprising:
Module is obtained, for obtaining facial image;
It cuts and gradation processing module, the facial image for obtaining to the acquisition module carries out cutting and gray scale
To obtain target signature region grayscale image, the target signature region grayscale image is the target signature area of the facial image for processing
The corresponding grayscale image in domain;
Computing module, for carrying out level side to each pixel in the target signature region grayscale image for obtaining module acquisition
To convolution algorithm and vertical direction convolution algorithm, to obtain the horizontal direction gradient and vertical gradient of each pixel;
First determining module, the horizontal direction gradient of each pixel for being calculated according to computing module and vertical
Direction gradient determines the corresponding approximate gradient of each pixel;
Second determining module, the corresponding approximate gradient of each pixel for being determined according to the first determining module determine mesh
Fuzziness is marked, using objective fuzzy degree as the fuzziness of facial image.
A kind of computer equipment, including memory, processor and storage are in the memory and can be in the processing
The computer program run on device, the processor realize that above-mentioned image blur determines method when executing the computer program
The step of.A kind of computer readable storage medium, the computer-readable recording medium storage have computer program, the calculating
Machine program realizes the step of above-mentioned image blur determines method when being executed by processor.
Above-mentioned image blur determines in the scheme that method, apparatus, computer equipment and storage medium are realized, for
It is no longer to be judged based on whole facial image, but obtained from facial image when the fuzziness of facial image is determined
Take target signature region to determine the fuzziness of facial image, while reducing calculation amount, due to be not take it is bulky
Opencv can also reduce the burden of the terminal device using the application, effectively improve the judgement of facial image fuzziness
Efficiency.
Detailed description of the invention
Technical solution in ord to more clearly illustrate embodiments of the present application, below by institute in the description to the embodiment of the present application
Attached drawing to be used is needed to be briefly described, it should be apparent that, the accompanying drawings in the following description is only some implementations of the application
Example, for those of ordinary skill in the art, without any creative labor, can also be according to these attached drawings
Obtain other attached drawings.
Fig. 1 is that image blur determines one application environment schematic diagram of method in the application;
Fig. 2 is that image blur determines method one embodiment flow diagram in the application;
Fig. 3 is mono- specific embodiment flow diagram of step S20 in Fig. 2;
Fig. 4 is mono- specific embodiment flow diagram of step S30 in Fig. 2;
Fig. 5 is another specific embodiment flow diagram of step S30 in Fig. 2;
Fig. 6 is mono- specific embodiment flow diagram of step S40 in Fig. 2;
Fig. 7 is another specific embodiment flow diagram of step S40 in Fig. 2;
Fig. 8 is image blur determining device one embodiment structural schematic diagram in the application;
Fig. 9 is an example structure schematic diagram of computer equipment in one embodiment of the application.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present application, technical solutions in the embodiments of the present application carries out clear, complete
Site preparation description, it is clear that described embodiment is some embodiments of the present application, instead of all the embodiments.Based on this Shen
Please in embodiment, every other implementation obtained by those of ordinary skill in the art without making creative efforts
Example, shall fall in the protection scope of this application.
The application provides a kind of image blur and determines method, and final fuzziness judging efficiency can be improved, can be applied to
In network frame schematic diagram as shown in Figure 1, wherein terminal device obtains facial image, carries out cutting and ash to facial image
Degree processing is to obtain target signature region grayscale image, wherein target signature region grayscale image is the target signature area of facial image
The corresponding grayscale image in domain;Horizontal direction convolution algorithm and vertical direction are carried out to pixel each in target signature region grayscale image
Convolution algorithm, to obtain the horizontal direction gradient and vertical gradient of each pixel;According to the level side of each pixel
The corresponding approximate gradient of each pixel is determined to gradient and vertical gradient;According to the corresponding approximate gradient of each pixel
Objective fuzzy degree is determined, using objective fuzzy degree as the fuzziness of facial image.Thus, in this application, for people
It is no longer to be judged based on whole facial image, but obtained from facial image when the fuzziness of face image is determined
Target signature region determines the fuzziness of facial image, while reducing calculation amount, due to be not take it is bulky
Opencv can also reduce the burden of the terminal device using the application, effectively improve the judgement of facial image fuzziness
Efficiency.Terminal device can be, but not limited to various personal computers, laptop, smart phone, tablet computer and portable
Wearable device is particularly suitable for mobile terminal device, such as mobile phone, wearable device etc., specifically without limitation.It additionally needs
Illustrate, the image blur in the application determines that method applies also in server, and server can use independent clothes
The server cluster of business device either multiple servers composition realizes, specifically without limitation.It is carried out below for the application detailed
Thin description, as shown in Fig. 2, including the following steps:
S10, facial image is obtained;
It is applied it is appreciated that face recognition technology is also widely used in every field, illustratively, for example is utilized
Recognition of face carries out mobile terminal or computer unlock, and the body of some application programs (Application, APP) is entered using mobile phone
Part verification process (such as shopping class APP, educational APP), using the process of mobile terminal progress attendance, in some gate inhibitions' monitoring
Recognition of face used etc., in above-mentioned face recognition process, the function of thering is brush face to verify, that is, need to obtain use
The facial image at family then carries out recognition of face.It should be noted that above-mentioned application example is that an example is said herein
It is bright, can't the application constitute and limit, specifically here without limitation, in addition, in a particular application, in addition to utilizing mobile phone, computer
Equal terminal devices can also be other terminal devices outside being unlocked, for example, smartwatch etc., here also without limitation.Upper
It states in exemplary application scenarios, in order to improve the accuracy of recognition of face, the facial image for the user that can be obtained first carries out one
The process of a Quality estimation, above-mentioned Quality estimation include the process that image blur determines, facial image has been obscured very much and will lead to
Last recognition of face effect can be poor, thus need first to judge before carrying out recognition of face the facial image obtained whether mould
Paste, until the facial image that taking-up meets quality is being identified.Therefore, in this application, above-mentioned facial image need to first be obtained.
S20, cutting and gray proces are carried out to facial image to obtain target signature region grayscale image, target signature region
Grayscale image is the corresponding grayscale image of target signature region of facial image;
In this application, after having got above-mentioned facial image, facial image need to be carried out cut and gray proces with
Obtain target signature region grayscale image.Wherein, the target signature region grayscale image is right for the target signature region of facial image
The grayscale image answered.In addition, facial image can be obtained in application scenes, the target signature region of facial image is cut out,
Grayscale image conversion is carried out to obtain target signature region grayscale image to target signature region.In other application scenarios,
Facial image can be obtained, grayscale image conversion is carried out to facial image, then cut from the facial image after grayscale image converts
Target signature region is out to obtain above-mentioned target signature region grayscale image, specifically here without limitation.Illustratively, it is carrying out
During above-mentioned grayscale image conversion, conventional grayscale image method for transformation can be used illustratively, can to make here without limitation
It is converted with RGB bitmap.For example, the color of certain original point is RGB (R, G, B) in facial image, it specifically can be by following several
By this, certain point is converted to obtain certain point Gray kind mode, comprising: 1, floating-point arithmetic: Gray=R*0.3+G*0.59+B*
0.11;2, integer method: Gray=(R*30+G*59+B*11)/100;3, displacement method: Gray=(R*76+G*151+B*
28)>>8;4, mean value method: Gray=(R+G+B)/3, specifically without limitation.
It is further to note that in some embodiments, after obtaining target signature region grayscale image, after improving
Continuous treatment effeciency and quality can pre-process target signature region grayscale image, wherein above-mentioned pretreatment includes but not office
It is limited to: denoising is carried out to target signature region grayscale image, that is, removes the point changed greatly in picture with surrounding point, with
And normalized, that is, the point for largely concentrating on centre in target signature region grayscale image is made to be stretched to entire grayscale image area
Domain.In denoising, illustratively, Gassian low-pass filter denoising specifically can be used, NL-Means (non-local mean) etc. is gone
It makes an uproar mode, here without limitation.And in normalized, common normalization means can be used, as long as can be by target signature
The pixel value normalizing of area grayscale figure is to a certain range, specifically without limitation, such as by target signature region grayscale image
Pixel value subtracts mean value, is then normalized divided by the mode of pixel value variance.It is appreciated that pass through above-mentioned pretreatment,
The interference information for can be very big reducing target signature region gray scale, effectively improves the overall treatment efficiency of scheme.
S30, horizontal direction convolution algorithm and vertical direction volume are carried out to pixel each in target signature region grayscale image
Product operation, to obtain the horizontal direction gradient and vertical gradient of each pixel;
It should be understood that target signature region grayscale image is made of multiple pixels, by obtaining target signature region ash
After spending figure, horizontal direction convolution algorithm and vertical direction convolution are carried out respectively to each pixel in target signature region grayscale image
Operation, so as to obtain the horizontal direction gradient and vertical gradient of each pixel.It should be understood that facial image is a kind of
Digital picture is a kind of discrete point spectrogram, and corresponding target signature region grayscale image is also a kind of spectrogram, is alternatively referred to as
A kind of two-dimensional discrete function, and gradient refers to the derivation of this two-dimensional discrete function, gradient has direction, and in this application,
Horizontal direction convolution algorithm and vertical direction convolution algorithm can be carried out to pixel each in target signature region grayscale image, to obtain
Each pixel horizontal direction gradient and vertical gradient, from another point of view on say, and to have obtained entire target special
Levy the horizontal direction gradient and vertical gradient of area grayscale figure.
S40, determine that each pixel is corresponding close according to the horizontal direction gradient and vertical gradient of each pixel
Like gradient;
In this application, in obtaining target signature region grayscale image one of target pixel points horizontal direction gradient
After vertical gradient, object pixel can be determined according to the horizontal direction gradient and vertical gradient of target pixel points
Put corresponding approximate gradient, other corresponding approximate gradient calculation methods of all pixels point and mesh in target signature region grayscale image
As marking pixel, therefore the corresponding approximate gradient of each pixel in target signature region grayscale image can be obtained.Wherein, as
The corresponding approximate gradient of vegetarian refreshments horizontal direction gradient corresponding to the pixel is related with vertical gradient.
S50, objective fuzzy degree is determined according to the corresponding approximate gradient of each pixel, using objective fuzzy degree as face figure
The fuzziness of picture.
Thus, being no longer based on whole people when the fuzziness for facial image is determined in this application
Face image is judged, but the fuzziness that target signature region determines facial image is obtained from facial image, is counted reducing
While calculation amount, due to not being to take bulky opencv, the burden of the terminal device using the application can also be reduced, is had
Improve to effect the judging efficiency of facial image fuzziness.
It should be noted that the target signature region of above-mentioned facial image refers to face figure in application scenes
The nasal area or eye areas of picture.In this application, the target signature region of facial image, which refers to, can most reflect people in face
Region of face notable feature, such as eyes, mouth, nose etc. are also to focus principally on the above-mentioned face area of identification in recognition of face
Domain, as long as above-mentioned target signature region fuzziness is enough, the fuzziness for reflecting facial image to a certain extent is enough, because
This is used as subsequent determination using the nasal area of above-mentioned facial image or eye areas as the target signature region in the application
The zoning of fuzziness.
In this application, objective fuzzy degree is being determined according to the corresponding approximate gradient of each pixel, by objective fuzzy degree
After the fuzziness of facial image, so that it may the quality of facial image is determined according to the image blur of facial image, to sentence
The fixed facial image is if appropriate for image used in the recognition of face as next step.
In one embodiment, as shown in figure 3, in step S20, namely to each pixel in target signature region grayscale image
Horizontal direction convolution algorithm and vertical direction convolution algorithm are carried out, to obtain the horizontal direction gradient and Vertical Square of each pixel
To gradient, specifically comprise the following steps:
S21, the horizontal warp factor of Sobel and the vertical warp factor of Sobel are determined;
S22, level side is carried out according to pixel each in the horizontal convolution factor pair target signature region grayscale image of Sobel
The horizontal direction gradient of each pixel is obtained to convolution algorithm, and according to the vertical warp factor of Sobel to target signature
Each pixel carries out vertical direction convolution algorithm to obtain the vertical gradient of each pixel in area grayscale figure.
It should be noted that in addition to calculating above-mentioned target signature region grayscale image according to the horizontal warp factor of Sobel
Horizontal direction gradient, and calculate according to the vertical warp factor of Sobel the vertical direction of above-mentioned target signature region grayscale image
Outside gradient direction, as long as pass that is adjacent in target signature region grayscale image or being separated by color difference between several pixels can be calculated
System, to calculate the horizontal direction gradient of the pixel of target signature region grayscale image or the warp factor of vertical gradient
It is ok, illustratively, in application scenes, the warp factor of usable Roberts (Luo Baici) convolution operator,
Warp factor, warp factor of Laplacian (Laplce Gauss) examination paper of Prewitt convolution operator etc., do not limit specifically
It is fixed.It is appreciated that in the present embodiment, specifically using Sobel Operator namely the horizontal warp factor of Sobel and rope
The vertical warp factor of Bell proposes the computational algorithm for carrying out horizontal direction gradient and vertical gradient, improves scheme
Exploitativeness.It is noted that can determine depositing to pixel side in target signature region using Sobel convolution operator
And pixel from secretly to bright, or the variation from bright to dark, therefore, the level calculated using Sobel convolution operator
Direction gradient or vertical gradient can be very square the fog-level for reflecting target signature region grayscale image so that last calculate
Objective fuzzy degree out can more reflect the Fuzzy Level of target signature region grayscale image, that is, make final facial image
The judgement of fuzziness is more accurate, therefore, in this application, it is preferable to use Sobel convolution operator.
In one embodiment, the horizontal warp factor of Sobel is [- 1,0,1, -2,0,2, -1,0,1], and Sobel is vertically rolled up
The product factor is [1,2,1,0,0,0, -1, -2, -1], namely:
The horizontal warp factor of the Sobel of Sobel convolution operator:
-1 | 0 | +1 |
-2 | 0 | +2 |
-1 | 0 | +1 |
The vertical warp factor of the Sobel of Sobel convolution operator are as follows:
+1 | +2 | +1 |
0 | 0 | 0 |
-1 | -2 | -1 |
Alternatively,
The horizontal warp factor of Sobel is [- 0.5,0,0.5, -1.5,0,1.5, -0.5,0,0.5], the vertical convolution of Sobel
The factor is [0.5,1.5,0.5,0,0,0, -0.5, -1.5, -0.5], namely:
The horizontal warp factor of the Sobel of Sobel convolution operator are as follows:
-0.5 | 0 | -1.5 |
-1.5 | 0 | 1.5 |
-0.5 | 0 | 0.5 |
The vertical warp factor of the Sobel of Sobel convolution operator are as follows:
0.5 | 0 | 1.5 |
0 | 0 | 0 |
-0.5 | -1.5 | -0.5 |
It is understood that in the present embodiment, propose the horizontal convolution of Sobel of two kinds of specific Sobel convolution operators because
Son warp factor vertical with Sobel, improves the exploitativeness of the application.Although but it should be recognized that above embodiment
In warp factor size be 3*3 template, but in some embodiments of the present application, the horizontal warp factor of Sobel or rope shellfish
You not also necessarily are 3*3 template at the size of vertical warp factor, illustratively, be also possible to 1*3 template, 2*4 template or
Other sizes of person, such as the horizontal warp factor of Sobel or the vertical warp factor of Sobel of 1*2 template or 2*2 template, specifically
Here without limitation.
In one embodiment, it as shown in figure 4, in step S30, namely according to the horizontal direction gradient of each pixel and hangs down
Straight direction gradient determines the corresponding approximate gradient of each pixel, specifically comprises the following steps:
S31, determine each pixel horizontal direction gradient absolute value and vertical gradient absolute value;
S32, it determines between the absolute value of horizontal direction gradient of each pixel and the absolute value of vertical gradient
With;
S33, by between the absolute value of horizontal direction gradient and the absolute value of vertical gradient of each pixel and
As the corresponding approximate gradient of each pixel.
It is, in the present embodiment, proposing a kind of side for specifically determining the corresponding approximate gradient of each pixel
Formula improves the exploitativeness of scheme, and above-mentioned calculating process, is described below by specific example in order to facilitate understanding, false
If target signature region grayscale image is A, the horizontal warp factor of Sobel is Gx, and the vertical warp factor of Sobel is Gy, then target
The horizontal direction gradient of characteristic area grayscale image all pixels point can be expressed as: GX=A*Gx, target signature region grayscale image institute
Having the vertical gradient of pixel can be expressed as: GY=A*Gy, then all pixels point of target signature region grayscale image A is close
It is indicated like the following formula of gradient:
G=| GX |+| GY |;
The corresponding approximate gradient of each pixel in target signature region grayscale image A namely can be obtained.
In one embodiment, it as shown in figure 5, in step S30, namely according to the horizontal direction gradient of each pixel and hangs down
Straight direction gradient determines the corresponding approximate gradient of each pixel, specifically comprises the following steps:
S31`, the horizontal direction gradient of each pixel and the quadratic sum of vertical gradient are determined;
S32`, root of making even to the horizontal direction gradient of each pixel and the quadratic sum of vertical gradient;
S33`, using the square root of each pixel as the corresponding approximate gradient of each pixel.
It is, in the present embodiment, proposing another kind and specifically determining the corresponding approximate gradient of each pixel
Mode, improves the exploitativeness of scheme, in order to make it easy to understand, being described below by specific example, it is assumed that target signature
Area grayscale figure is A, and the horizontal warp factor of Sobel is Gx, and the vertical warp factor of Sobel is Gy, then target signature region ash
The horizontal direction gradient for spending figure all pixels point is GX=A*Gx, the Vertical Square of target signature region grayscale image all pixels point
It is GY=A*Gy to gradient, then the following formula of the approximate gradient of all pixels point of target signature region grayscale image A indicates:
The corresponding approximate gradient of each pixel in target signature region grayscale image A namely can be obtained.
It should be noted that being determined often in addition to above-mentioned according to the horizontal direction gradient and vertical gradient of each pixel
Outside the mode of the corresponding approximate gradient of a pixel, there can also be other calculations, specifically without limitation, for example, by every
The approximate gradient of the horizontal direction gradient of a pixel and the arithmetic mean of instantaneous value of vertical gradient as pixel, to target spy
The calculation of each pixel takes the calculation of above-mentioned arithmetic mean of instantaneous value in sign area grayscale figure, to obtain each
The corresponding approximate gradient of pixel.It is appreciated that horizontal direction gradient with specific reference to each pixel set forth above and hanging down
Straight direction gradient determines the mode of the corresponding approximate gradient of each pixel, improves the exploitativeness of scheme.
In one embodiment, as shown in fig. 6, in step S40, target is determined according to the corresponding approximate gradient of each pixel
Fuzziness specifically comprises the following steps;
S41, the sum of the corresponding degree of approximation of each pixel is determined;
S42, by the sum of corresponding degree of approximation of each pixel divided by target signature region grayscale image pixel sum, with
Obtain objective fuzzy degree.
Here, a kind of mode that objective fuzzy degree is determined with specific reference to the corresponding approximate gradient of each pixel is proposed,
Improve the exploitativeness of scheme.In addition, since the corresponding approximate gradient of each pixel has reacted face to a certain extent
Therefore the fuzziness of image determines that the fuzziness of facial image can be used for judging according to the corresponding approximate gradient of each pixel
The fog-level of facial image, one of the reference as quality of human face image.
In one embodiment, as shown in fig. 7, before step S40, mesh is determined according to the corresponding approximate gradient of each pixel
Before marking fuzziness, method further includes following steps: calculating the brightness value of all pixels point in the facial image;Described in calculating
The corresponding natural logrithm of brightness value of all pixels point in facial image;It is corresponding to the brightness value of all pixels point in facial image
Natural logrithm average;Right index value is derived to obtain the average brightness of the facial image to the average value.Specifically
Shown in following formula: assuming that each brightness value is indicated with l (x) in facial image, then the average brightness of the facial image can be by
Following formula is calculated:Wherein, δ is preset value, can be 0.0001, tool
Body is without limitation.
In step S40, namely objective fuzzy degree determined according to the corresponding approximate gradient of each pixel, specifically included as follows
Step;
S41`, determine that the target for being greater than the average brightness of facial image in the corresponding approximate gradient of each pixel is approximate
Degree;
S42`, by the sum of target degree of approximation divided by target signature region grayscale image pixel sum, to obtain target mould
Paste degree.
In the present embodiment, before determining objective fuzzy degree according to the corresponding approximate gradient of each pixel, people is also utilized
The average brightness of face image has comprehensively considered the brightness and approximate gradient of facial image, can further improve face as threshold value
The accuracy for the fuzziness that image is determined.
It should be understood that the size of the serial number of each step is not meant that the order of the execution order in above-described embodiment, each process
Execution sequence should be determined by its function and internal logic, the implementation process without coping with the embodiment of the present application constitutes any limit
It is fixed.
In one embodiment, a kind of image blur determining device is provided, the image blur determining device and above-mentioned reality
It applies image blur in example and determines that method corresponds.As shown in figure 8, the image blur determining device 80 includes obtaining module
801, cutting and gradation processing module 802, computing module 803, the first determining module 804 and the second determining module 805.Each function
Detailed description are as follows for module:
Module 801 is obtained, for obtaining facial image;
It cuts and gradation processing module 802, the facial image for obtaining to acquisition module 801 carries out at cutting and gray scale
For reason to obtain target signature region grayscale image, target signature region grayscale image is the corresponding ash of target signature region of facial image
Degree figure;
Computing module 803, in the target signature region grayscale image to cutting and the acquisition of gradation processing module 802
Each pixel carries out horizontal direction convolution algorithm and vertical direction convolution algorithm, to obtain the level side of each pixel
To gradient and vertical gradient;
First determining module 804, the level of each pixel for being calculated according to the computing module 803
Direction gradient and vertical gradient determine the corresponding approximate gradient of each pixel;
Second determining module 805, each pixel for being determined according to first determining module 804 are corresponding
Approximate gradient determines objective fuzzy degree, using the objective fuzzy degree as the fuzziness of the facial image.
In one embodiment, computing module 803 is specifically used for:
Determine the horizontal warp factor of Sobel and the vertical warp factor of Sobel;
Horizontal direction volume is carried out according to pixel each in the horizontal convolution factor pair target signature region grayscale image of Sobel
Operation is accumulated to obtain the horizontal direction gradient of each pixel, and according to the vertical warp factor of Sobel to target signature region
Each pixel carries out vertical direction convolution algorithm to obtain the vertical gradient of each pixel in grayscale image.
In one embodiment,
The horizontal warp factor of Sobel be [- 1,0,1, -2,0,2, -1,0,1], the vertical warp factor of Sobel be [1,2,
1,0,0,0, -1, -2, -1];Namely:
The horizontal warp factor of Sobel:
-1 | 0 | +1 |
-2 | 0 | +2 |
-1 | 0 | +1 |
The vertical warp factor of the Sobel of Sobel convolution operator are as follows:
+1 | +2 | +1 |
0 | 0 | 0 |
-1 | -2 | -1 |
Alternatively,
The horizontal warp factor of Sobel is [- 0.5,0,0.5, -1.5,0,1.5, -0.5,0,0.5], the vertical convolution of Sobel
The factor is [0.5,1.5,0.5,0,0,0, -0.5, -1.5, -0.5], namely:
The horizontal warp factor of the Sobel of Sobel convolution operator are as follows:
The vertical warp factor of the Sobel of Sobel convolution operator are as follows:
0.5 | 0 | 1.5 |
0 | 0 | 0 |
-0.5 | -1.5 | -0.5 |
In one embodiment, the first determining module 804, is specifically used for:
Determine the absolute value of the horizontal direction gradient of each pixel and the absolute value of vertical gradient;
Determine it is between the absolute value of horizontal direction gradient of each pixel and the absolute value of vertical gradient and;
By between the absolute value of horizontal direction gradient and the absolute value of vertical gradient of each pixel and conduct
The corresponding approximate gradient of each pixel.
In one embodiment, the first determining module 804, is specifically used for:
Determine the horizontal direction gradient of each pixel and the quadratic sum of vertical gradient;
The quadratic sum of horizontal direction gradient and vertical gradient to each pixel is made even root;
Using the square root of each pixel as the corresponding approximate gradient of each pixel.
In one embodiment, the second determining module 805, is specifically used for;
Determine the sum of corresponding degree of approximation of each pixel;
By the sum of corresponding degree of approximation of each pixel divided by the pixel sum of target signature region grayscale image, to obtain
Objective fuzzy degree.
In one embodiment, the second determining module 805, is specifically used for:
Before determining objective fuzzy degree according to the corresponding approximate gradient of each pixel, calculates in the facial image and own
The brightness value of pixel;Calculate the corresponding natural logrithm of brightness value of all pixels point in the facial image;To facial image
The corresponding natural logrithm of brightness value of middle all pixels point is averaged;It is described to obtain to be derived from right index value to the average value
The average brightness of facial image;
It determines in the corresponding approximate gradient of each pixel greater than the target degree of approximation of the average brightness of facial image;
By the sum of target degree of approximation divided by the pixel sum of target signature region grayscale image, to obtain objective fuzzy degree.
Thus, present applicant proposes a kind of image blur determining device, the fuzziness for facial image into
It is no longer to be judged based on whole facial image, but obtain target signature region from facial image and determine when row determines
The fuzziness of facial image, due to not being to take bulky opencv, can also reduce use while reducing calculation amount
The burden of the terminal device of the application effectively improves the judging efficiency of facial image fuzziness.
Specific restriction about image blur determining device may refer to determine method above for image blur
Restriction, details are not described herein.Modules in above-mentioned image blur determining device can be fully or partially through software, hard
Part and combinations thereof is realized.Above-mentioned each module can be embedded in the form of hardware or independently of in the processor in computer equipment,
It can also be stored in a software form in the memory in computer equipment, execute the above modules in order to which processor calls
Corresponding operation.
In one embodiment, a kind of computer equipment is provided, which can be terminal, internal structure
Figure can be as shown in Figure 9.The computer equipment includes processor, the memory, network interface, display connected by system bus
Screen and input unit.Wherein, the processor of the computer equipment is for providing calculating and control ability.The computer equipment is deposited
Reservoir includes non-volatile memory medium, built-in storage.The non-volatile memory medium is stored with operating system and computer journey
Sequence.The built-in storage provides environment for the operation of operating system and computer program in non-volatile memory medium.The calculating
To realize that a kind of image blur determines method when machine program is executed by processor.
In one embodiment, a kind of computer equipment is provided, including memory, processor and storage are on a memory
And the computer program that can be run on a processor, processor perform the steps of when executing computer program
Obtain facial image;
Facial image cut and gray proces are to obtain target signature region grayscale image, target signature region gray scale
Figure is the corresponding grayscale image of target signature region of facial image;
Horizontal direction convolution algorithm and vertical direction convolution fortune are carried out to pixel each in target signature region grayscale image
It calculates, to obtain the horizontal direction gradient and vertical gradient of each pixel;
The corresponding approximate ladder of each pixel is determined according to the horizontal direction gradient and vertical gradient of each pixel
Degree;
Objective fuzzy degree is determined according to the corresponding approximate gradient of each pixel, using objective fuzzy degree as facial image
Fuzziness.
Thus, present applicant proposes a kind of computer equipment, when the fuzziness for facial image is determined,
It is no longer to be judged based on whole facial image, but obtain target signature region from facial image and determine facial image
Fuzziness, due to not being to take bulky opencv, can also reduce the end using the application while reducing calculation amount
The burden of end equipment effectively improves the judging efficiency of facial image fuzziness, it is further to note that the application provides
Computer equipment in, the step of realizing or function can be corresponded to refering to preceding method embodiment when processor executes computer program
In description, do not repeat to repeat here.
In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated
Machine program performs the steps of when being executed by processor
Obtain facial image;
Facial image cut and gray proces are to obtain target signature region grayscale image, target signature region gray scale
Figure is the corresponding grayscale image of target signature region of facial image;
Horizontal direction convolution algorithm and vertical direction convolution fortune are carried out to pixel each in target signature region grayscale image
It calculates, to obtain the horizontal direction gradient and vertical gradient of each pixel;
The corresponding approximate ladder of each pixel is determined according to the horizontal direction gradient and vertical gradient of each pixel
Degree;
Objective fuzzy degree is determined according to the corresponding approximate gradient of each pixel, using objective fuzzy degree as facial image
Fuzziness.
Thus, present applicant proposes a kind of computer readable storage medium, the fuzziness for facial image into
It is no longer to be judged based on whole facial image, but obtain target signature region from facial image and determine when row determines
The fuzziness of facial image, due to not being to take bulky opencv, can also reduce use while reducing calculation amount
The burden of the terminal device of the application effectively improves the judging efficiency of facial image fuzziness, it is further to note that
In computer readable storage medium provided by the present application, the step of computer program is realized when being executed by processor or function,
It can correspond to refering to the description in preceding method embodiment, not repeat to repeat here.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, the computer program can be stored in a non-volatile computer
In read/write memory medium, the computer program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein,
To any reference of memory, storage, database or other media used in each embodiment provided herein,
Including non-volatile and/or volatile memory.Nonvolatile memory may include read-only memory (ROM), programming ROM
(PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include
Random access memory (RAM) or external cache.By way of illustration and not limitation, RAM is available in many forms,
Such as static state RAM (SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM), enhancing
Type SDRAM (ESDRAM), synchronization link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM
(RDRAM), direct memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
It is apparent to those skilled in the art that for convenience of description and succinctly, only with above-mentioned each function
Can unit, module division progress for example, in practical application, can according to need and by above-mentioned function distribution by different
Functional unit, module are completed, i.e., the internal structure of described device is divided into different functional unit or module, more than completing
The all or part of function of description.
Embodiment described above is only to illustrate the technical solution of the application, rather than its limitations;Although referring to aforementioned reality
Example is applied the application is described in detail, those skilled in the art should understand that: it still can be to aforementioned each
Technical solution documented by embodiment is modified or equivalent replacement of some of the technical features;And these are modified
Or replacement, the spirit and scope of each embodiment technical solution of the application that it does not separate the essence of the corresponding technical solution should all
Comprising within the scope of protection of this application.
Claims (10)
1. a kind of image blur determines method characterized by comprising
Obtain facial image;
The facial image cut and gray proces are to obtain target signature region grayscale image, the target signature region
Grayscale image is the corresponding grayscale image of target signature region of the facial image;
Horizontal direction convolution algorithm and vertical direction convolution fortune are carried out to pixel each in the target signature region grayscale image
It calculates, to obtain the horizontal direction gradient and vertical gradient of each pixel;
Determine that each pixel is corresponding close according to the horizontal direction gradient of each pixel and vertical gradient
Like gradient;
Objective fuzzy degree is determined according to the corresponding approximate gradient of each pixel, using the objective fuzzy degree as the people
The fuzziness of face image.
2. image blur as described in claim 1 determines method, which is characterized in that described to the target signature region ash
It spends each pixel in figure and carries out horizontal direction convolution algorithm and vertical direction convolution algorithm, to obtain each pixel
Horizontal direction gradient and vertical gradient, comprising:
Determine the horizontal warp factor of Sobel and the vertical warp factor of Sobel;
Each pixel according to target signature region grayscale image described in the horizontal convolution factor pair of the Sobel carries out water
Square obtain the horizontal direction gradient of each pixel to convolution algorithm, and according to the vertical convolution of the Sobel because
It is described each to obtain that son carries out vertical direction convolution algorithm to each pixel described in the target signature region grayscale image
The vertical gradient of pixel.
3. image blur as claimed in claim 2 determines method, which is characterized in that
The horizontal warp factor of Sobel is [- 1,0,1, -2,0,2, -1,0,1], and the vertical warp factor of the Sobel is
[1,2,1,0,0,0, -1, -2, -1];
Or,
The horizontal warp factor of Sobel is [- 0.5,0,0.5, -1.5,0,1.5, -0.5,0,0.5], and the Sobel is vertical
Warp factor is [0.5,1.5,0.5,0,0,0, -0.5, -1.5, -0.5].
4. image blur as claimed in claim 3 determines method, which is characterized in that described according to each pixel
Horizontal direction gradient and vertical gradient determine the corresponding approximate gradient of each pixel, comprising:
Determine the absolute value of the horizontal direction gradient of each pixel and the absolute value of vertical gradient;
Determine it is between the absolute value of horizontal direction gradient and the absolute value of vertical gradient of each pixel and;
By between the absolute value of horizontal direction gradient and the absolute value of vertical gradient of each pixel and conduct
The corresponding approximate gradient of each pixel.
5. image blur as claimed in claim 3 determines method, which is characterized in that described according to each pixel
Horizontal direction gradient and vertical gradient determine the corresponding approximate gradient of each pixel, comprising:
Determine the horizontal direction gradient of each pixel and the quadratic sum of vertical gradient;
The quadratic sum of horizontal direction gradient and vertical gradient to each pixel is made even root;
Using the square root of each pixel as the corresponding approximate gradient of each pixel.
6. image blur as described in claim 1 determines method, which is characterized in that described according to each pixel pair
The approximate gradient answered determines objective fuzzy degree, including;
Determine the sum of corresponding degree of approximation of each pixel;
By the sum of corresponding degree of approximation of each pixel divided by the pixel sum of the target signature region grayscale image, with
Obtain the objective fuzzy degree.
7. image blur as described in claim 1 determines method, which is characterized in that described according to each pixel pair
Before the approximate gradient answered determines objective fuzzy degree, the method also includes:
Calculate the brightness value of all pixels point in the facial image;
Calculate the corresponding natural logrithm of brightness value of all pixels point in the facial image;
It averages to the corresponding natural logrithm of brightness value of all pixels point in facial image;
Right index value is derived to obtain the average brightness of the facial image to the average value;
It is described that objective fuzzy degree is determined according to the corresponding approximate gradient of each pixel, including;
It determines approximate greater than the target of the average brightness of the facial image in the corresponding approximate gradient of each pixel
Degree;
By the sum of described target degree of approximation divided by the pixel sum of the target signature region grayscale image, to obtain the target
Fuzziness.
8. a kind of image blur determining device characterized by comprising
Module is obtained, for obtaining facial image;
It cuts and gradation processing module, the facial image for obtaining to the acquisition module carries out cutting and gray proces
To obtain target signature region grayscale image, the target signature region grayscale image is the target signature region pair of the facial image
The grayscale image answered;
Computing module, for every in the cutting and the target signature region grayscale image that handles of gradation processing module
A pixel carries out horizontal direction convolution algorithm and vertical direction convolution algorithm, to obtain the horizontal direction of each pixel
Gradient and vertical gradient;
First determining module, the horizontal direction gradient of each pixel for being calculated according to the computing module and
Vertical gradient determines the corresponding approximate gradient of each pixel;
Second determining module, the corresponding approximate gradient of each pixel for being determined according to first determining module are true
Set the goal fuzziness, using the objective fuzzy degree as the fuzziness of the facial image.
9. a kind of computer equipment, including memory, processor and storage are in the memory and can be in the processor
The computer program of upper operation, which is characterized in that the processor realized when executing the computer program as claim 1 to
The step of any one of 7 described image fuzzinesses determine method.
10. a kind of computer readable storage medium, the computer-readable recording medium storage has computer program, and feature exists
In any one of such as claim 1 to 7 of realization described image fuzziness determines method when the computer program is executed by processor
The step of.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811073282.2A CN109447942B (en) | 2018-09-14 | Image ambiguity determining method, apparatus, computer device and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811073282.2A CN109447942B (en) | 2018-09-14 | Image ambiguity determining method, apparatus, computer device and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109447942A true CN109447942A (en) | 2019-03-08 |
CN109447942B CN109447942B (en) | 2024-04-23 |
Family
ID=
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110032964A (en) * | 2019-04-08 | 2019-07-19 | 腾讯科技(成都)有限公司 | Image processing method, method, apparatus, equipment and the storage medium for identifying visual angle |
CN113592720A (en) * | 2021-09-26 | 2021-11-02 | 腾讯科技(深圳)有限公司 | Image scaling processing method, device, equipment, storage medium and program product |
WO2022042131A1 (en) * | 2020-08-28 | 2022-03-03 | 稿定(厦门)科技有限公司 | Adaptive halo image generation method and apparatus based on particles |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2012073703A (en) * | 2010-09-28 | 2012-04-12 | Nippon Hoso Kyokai <Nhk> | Image blur amount calculation device and program for the same |
CN103269439A (en) * | 2013-05-21 | 2013-08-28 | 杭州电子科技大学 | OCT image quality objective non-reference type evaluation method |
CN103353985A (en) * | 2013-03-27 | 2013-10-16 | 西华大学 | Measurement method for image Gaussian Blur |
CN105488511A (en) * | 2015-11-25 | 2016-04-13 | 小米科技有限责任公司 | Image identification method and device |
CN105894506A (en) * | 2016-03-30 | 2016-08-24 | 深圳云天励飞技术有限公司 | Face image fuzziness computing method and device |
CN107580209A (en) * | 2017-10-24 | 2018-01-12 | 维沃移动通信有限公司 | Take pictures imaging method and the device of a kind of mobile terminal |
CN107767358A (en) * | 2016-08-23 | 2018-03-06 | 阿里巴巴集团控股有限公司 | A kind of objects in images fuzziness determines method and apparatus |
CN108234882A (en) * | 2018-02-11 | 2018-06-29 | 维沃移动通信有限公司 | A kind of image weakening method and mobile terminal |
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2012073703A (en) * | 2010-09-28 | 2012-04-12 | Nippon Hoso Kyokai <Nhk> | Image blur amount calculation device and program for the same |
CN103353985A (en) * | 2013-03-27 | 2013-10-16 | 西华大学 | Measurement method for image Gaussian Blur |
CN103269439A (en) * | 2013-05-21 | 2013-08-28 | 杭州电子科技大学 | OCT image quality objective non-reference type evaluation method |
CN105488511A (en) * | 2015-11-25 | 2016-04-13 | 小米科技有限责任公司 | Image identification method and device |
CN105894506A (en) * | 2016-03-30 | 2016-08-24 | 深圳云天励飞技术有限公司 | Face image fuzziness computing method and device |
CN107767358A (en) * | 2016-08-23 | 2018-03-06 | 阿里巴巴集团控股有限公司 | A kind of objects in images fuzziness determines method and apparatus |
CN107580209A (en) * | 2017-10-24 | 2018-01-12 | 维沃移动通信有限公司 | Take pictures imaging method and the device of a kind of mobile terminal |
CN108234882A (en) * | 2018-02-11 | 2018-06-29 | 维沃移动通信有限公司 | A kind of image weakening method and mobile terminal |
Non-Patent Citations (1)
Title |
---|
路晓亚 等: "模糊生物图像特征优化提取仿真研究", 计算机仿真, no. 05 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110032964A (en) * | 2019-04-08 | 2019-07-19 | 腾讯科技(成都)有限公司 | Image processing method, method, apparatus, equipment and the storage medium for identifying visual angle |
CN110032964B (en) * | 2019-04-08 | 2023-07-14 | 腾讯科技(成都)有限公司 | Image processing method, device, equipment and storage medium for identifying visual angle |
WO2022042131A1 (en) * | 2020-08-28 | 2022-03-03 | 稿定(厦门)科技有限公司 | Adaptive halo image generation method and apparatus based on particles |
CN113592720A (en) * | 2021-09-26 | 2021-11-02 | 腾讯科技(深圳)有限公司 | Image scaling processing method, device, equipment, storage medium and program product |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110826519B (en) | Face shielding detection method and device, computer equipment and storage medium | |
US10963993B2 (en) | Image noise intensity estimation method, image noise intensity estimation device, and image recognition device | |
CN111741211B (en) | Image display method and apparatus | |
CN105917353B (en) | Feature extraction and matching for biological identification and template renewal | |
CN106803067B (en) | Method and device for evaluating quality of face image | |
CN103914676B (en) | A kind of method and apparatus used in recognition of face | |
CN111274916B (en) | Face recognition method and face recognition device | |
CN108428214B (en) | Image processing method and device | |
JP4692856B2 (en) | Method and apparatus for model-based anisotropic diffusion | |
WO2014168587A1 (en) | Method and system for processing an input image | |
CN103902958A (en) | Method for face recognition | |
Kumar | Deblurring of motion blurred images using histogram of oriented gradients and geometric moments | |
CN110147708B (en) | Image data processing method and related device | |
CN110458792B (en) | Method and device for evaluating quality of face image | |
CN111079764A (en) | Low-illumination license plate image recognition method and device based on deep learning | |
CN109508636A (en) | Vehicle attribute recognition methods, device, storage medium and electronic equipment | |
WO2019061659A1 (en) | Method and device for removing eyeglasses from facial image, and storage medium | |
CN107256543A (en) | Image processing method, device, electronic equipment and storage medium | |
CN111222433A (en) | Automatic face auditing method, system, equipment and readable storage medium | |
CN113592776A (en) | Image processing method and device, electronic device and storage medium | |
CN110837781B (en) | Face recognition method, face recognition device and electronic equipment | |
CN113096023A (en) | Neural network training method, image processing method and device, and storage medium | |
CN111274851A (en) | Living body detection method and device | |
Fathy et al. | Benchmarking of pre-processing methods employed in facial image analysis | |
Han et al. | Locally adaptive contrast enhancement using convolutional neural network |
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 |