CN101950356A - Smiling face detecting method and system - Google Patents
Smiling face detecting method and system Download PDFInfo
- Publication number
- CN101950356A CN101950356A CN 201010276313 CN201010276313A CN101950356A CN 101950356 A CN101950356 A CN 101950356A CN 201010276313 CN201010276313 CN 201010276313 CN 201010276313 A CN201010276313 A CN 201010276313A CN 101950356 A CN101950356 A CN 101950356A
- Authority
- CN
- China
- Prior art keywords
- weak classifier
- smiling face
- submodule
- threshold value
- face image
- 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 abstract description 29
- 238000001514 detection method Methods 0.000 claims abstract description 61
- 238000012545 processing Methods 0.000 claims abstract description 18
- 238000011946 reduction process Methods 0.000 claims description 16
- 238000012549 training Methods 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 9
- 230000001105 regulatory effect Effects 0.000 claims description 9
- 230000007246 mechanism Effects 0.000 claims description 8
- 238000004800 variational method Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 abstract description 12
- 230000008030 elimination Effects 0.000 abstract 1
- 238000003379 elimination reaction Methods 0.000 abstract 1
- 238000005516 engineering process Methods 0.000 description 6
- 238000012360 testing method Methods 0.000 description 4
- 238000011161 development Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000001143 conditioned effect Effects 0.000 description 2
- 241001270131 Agaricus moelleri Species 0.000 description 1
- 241000272525 Anas platyrhynchos Species 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000001815 facial effect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 239000011521 glass Substances 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000452 restraining effect Effects 0.000 description 1
- 235000014347 soups Nutrition 0.000 description 1
- 230000026676 system process Effects 0.000 description 1
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention relates to computer image processing and discloses a smiling face detecting method and a smiling face detecting system. In the invention, an Adabost voting principle is introduced to a smiling face detecting algorithm to reduce the computing complexity in a detection process considerably, and thus, real-time smiling face detection can be realized on portable equipment with low computing capacity and low storing capacity. In addition, before the Adaboost algorithm is used for smiling face detection, the original face image is preprocessed by noise elimination and brightness regulation, the drawbacks of the use of the Adaboost in smiling face detection are overcome, the computing complexity in a detection process is reduced, and the accuracy of the detection is ensured.
Description
Technical field
The present invention relates to Computer Image Processing, particularly smiling face's detection technique of image.
Background technology
In recent years, be accompanied by Development of Multimedia Technology, also comprised increasing multimedia software in the portable sets such as mobile phone, digital camera, such as picture denoising, recognition of face etc., these application software are well received by consumers.Along with the improvement and the development of portable set, the smiling face detects the technology that also becomes an active demand in addition.
So-called smiling face detects, and just is meant whether to given human face region, detecting this people's face with specific method is the smiling face after finishing people's face and detecting, and process and method that the smiling face is marked.The smiling face detects and has very important significance, and as the automatic candid photograph of digital camera to the smiling face, the smile service of modern service industry etc. can both be used smiling face's recognition technology.But the smiling face detects not a duck soup, and its implementation procedure is faced with a lot of challenges, and for example, because the fuzzy behaviour of human expression, sometimes for a kind of fixedly expression, the judgement that different people provides may differ greatly.Owing to facial pose, appearance, the colour of skin, whether many condition influence such as shelter such as glasses and optical imagery environment are arranged, detection has been caused very big difficulty in addition.Influencing the smiling face, to detect two restraining factors moving towards practical application be accuracy and the speed that the smiling face detects.
Along with the development of computer image processing technology, the smiling face detection method that adopts has had tangible lifting in accuracy at present.But the present inventor finds that existing smiling face detection method complexity is all very high, and portable set exists usually problems such as computing power is poor, storage capacity is weak, therefore existing smiling face detection method can't be applied in these equipment well.Though existing Adaboost algorithm possesses the low characteristics of complexity, has solved speed issue well.But this method is used for the smiling face and detects and also to be faced with a lot of problems: be difficult to this method identification etc. as, smiling face very sensitive to noise and non-smiling face's difference is small, causing at present also can't be with the application in the smiling face detects of Adaboost algorithm.
Summary of the invention
The object of the present invention is to provide a kind of smiling face detection method and system thereof, make in real time, efficiently the smiling face detect can be poor in computing power, realize on the portable set a little less than the storage capacity.
For solving the problems of the technologies described above, embodiments of the present invention provide a kind of smiling face detection method, comprise following steps:
Original face image is carried out pre-service, and pre-service comprises denoising and brightness regulation processing;
Will be through pretreated face image, carry out smiling face's detection with the Adaboost algorithm of voting mechanism;
The result that the output smiling face detects.
Embodiments of the present invention also provide a kind of smiling face's detection system, comprise:
Pretreatment module is used for original face image is carried out pre-service, and pre-service comprises denoising and brightness regulation processing;
Detection module is used for carrying out smiling face's detection with through the pretreated face image of pretreatment module with the Adaboost algorithm;
Output module is used to export the result that smiling face that detection module carries out detects.
Embodiment of the present invention compared with prior art, the key distinction and effect thereof are:
Original face image is comprised denoising and brightness regulation handle in interior pre-service, will be through pretreated face image, carry out smiling face's detection with the Adaboost algorithm of voting mechanism, and the result that detects of output smiling face.Because the complexity of Adaboost algorithm is lower, therefore the principle of Adaboost being voted is incorporated in smiling face's detection algorithm, can make the computation complexity of testing process reduce significantly, thus can be poor, realize that real-time smiling face detects on the portable set a little less than the storage capacity in computing power.And, because it is very sensitive to noise based on the Adaboost algorithm of voting mechanism, to the difficult identification of smiling face and non-smiling face's a shade of difference, therefore before employing Adaboost algorithm carries out smiling face's detection, by denoising and brightness regulation, original face image has been carried out pre-service earlier, the insufficient section that Adaboost is applied to smiling face's identification improves, make and in reducing testing process, in the computation complexity, also can guarantee the accuracy that detects.
Further, when original face image is gone pre-service, carry out denoising earlier, carry out brightness regulation again and handle, reducing to greatest extent, thereby further guaranteed the accuracy that the smiling face detects the disadvantageous various factors of algorithm.
Further, in the mode of iteration face image is carried out noise reduction process one after another, the face image after noise reduction process and the difference of the face image before the noise reduction process satisfy pre-conditioned.By denoising method original face image is handled, can effectively be guaranteed denoising effect original face image based on the variational method.
Further, in the Adaboost algorithm, the Weak Classifier that is used to vote be h (x, f, p, θ),
Wherein, f (x) is the eigenwert of Weak Classifier, and f (x) gets the haar eigenwert, θ is used to judge whether the eigenwert of Weak Classifier satisfies the predetermined threshold value of smiling face's condition, and p is used to adjust the direction of the sign of inequality, as f (x) during less than threshold value θ, p=1, as f (x) during greater than threshold value θ, p=-1; The ballot of representing Weak Classifier during h=1 is the smiling face, and the ballot of representing Weak Classifier during h=0 is non-smiling face.Realize simply, further guaranteed the low complex degree that the smiling face detects.
Description of drawings
Fig. 1 is the smiling face detection method process flow diagram according to first embodiment of the invention;
Fig. 2 is according to the denoising process flow diagram in the first embodiment of the invention;
Fig. 3 is according to the brightness regulation processing flow chart in the first embodiment of the invention;
Fig. 4 is the smiling face's detection system process flow diagram according to third embodiment of the invention.
Embodiment
In the following description, in order to make the reader understand the application better many ins and outs have been proposed.But, persons of ordinary skill in the art may appreciate that even without these ins and outs with based on the many variations and the modification of following each embodiment, also can realize each claim of the application technical scheme required for protection.
For making the purpose, technical solutions and advantages of the present invention clearer, embodiments of the present invention are described in further detail below in conjunction with accompanying drawing.
First embodiment of the invention relates to a kind of smiling face detection method.Fig. 1 is the schematic flow sheet of this smiling face detection method.
In step 101, original face image is carried out pre-service, wherein, pre-service comprises denoising and brightness regulation processing.In this step, earlier original face image is carried out denoising, again the face image after denoising is carried out brightness regulation and handle.When original face image is gone pre-service, carry out denoising earlier, carry out brightness regulation again and handle, can further guarantee the accuracy that the smiling face detects.
Specifically, the flow process that original face image is carried out denoising as shown in Figure 2.In step 201, i makes zero with counter, and with asking denoising image u
iBe initialized as u
0=u
0, u
0=u
0Be original face image.
Then, in step 202, calculate the derivative of each pixel on line direction in the face image.Specifically, calculate u by following formula
iIn the derivative u of each pixel on line direction
i x(x, y):
Wherein,
The elements of a fix of x and y remarked pixel, the height of h presentation video.
Then, in step 203, calculate the derivative of each pixel on column direction in the face image.Specifically, calculate u by following formula
iIn the derivative u of each pixel on column direction
i y(x, y):
Wherein,
The elements of a fix of x and y remarked pixel, the width of w presentation video.
Then, in step 204,, calculate the Grad of each pixel according to the derivative of each pixel on line direction and column direction that calculates.Promptly calculate u according to following formula
iIn the Grad of each pixel:
Wherein, i, j represent the vector of unit length of ranks direction, u respectively
i x(x y) tries to achieve in step 202, u
i y(x y) tries to achieve in step 203.
Then, in step 205, according to the derivative of each pixel on line direction and column direction that calculates, the gradient-norm of trying to achieve each pixel of calculation is long.Promptly calculate u according to following formula
iIn the gradient-norm of each pixel long:
Wherein, u
i x(x y) tries to achieve in step 202, u
i y(x y) tries to achieve in step 203.Then, long according to the Grad and the gradient-norm of each pixel in step 206, try to achieve the divergence of the ratio of the Grad of each pixel and gradient-norm length.Promptly calculate according to following formula
Divergence:
Wherein, outer small tenon x, the y of equation the right bracket, the partial derivative of expression x and y direction.
Then, in step 207, according to following formula to face image u
iCarry out noise reduction process one time:
Wherein, u
I+1Expression is to u
iCarry out the face image after the noise reduction process one time, div
Represent the divergence of the ratio of the Grad of each pixel and gradient-norm length, t is the algorithm step-length that is used to regulate algorithm speed, and λ is default adjustable parameter, and the more little expression of λ requires image smooth more.T=λ=0.01 acquiescently.In addition, be appreciated that in actual applications that t and λ can also be made as other values, t and λ can be identical, also can be different, do not exemplify one by one at this.
Then, in step 208,, judge for given threshold epsilon (for example ε=0.1) || u
I+1-u
i|| whether＜ε sets up, if, then with u
I+1As the face image after denoising, withdraw from the flow process of denoising; If || u
I+1-u
i||＜ε is false, and then enters step 209.
In step 209, i adds up with counter, and promptly i=i+1 comes back to step 202.
Mode with iteration is carried out noise reduction process one after another to face image, that the face image after noise reduction process and the difference of the face image before the noise reduction process satisfy is pre-conditioned (|| u
I+1-u
i||＜ε).By denoising method original face image is handled, can effectively be guaranteed denoising effect original face image based on the variational method.
Face image after denoising is carried out flow process that brightness regulation handles as shown in Figure 3.
In step 301, calculate the average of the gray scale of all pixels in the face image to be regulated, face image to be regulated is the face image after denoising, represents with u.That is to say, obtain the average (w, h represent the width and the height of face image respectively) of all gray scales among the u:
Then, in step 302,,, adjust to a definite value with the gray scale of all pixels in the face image to be regulated according to the average of the gray scale of all pixels of calculating.The gray average that is about to u is transferred to definite value a, for example an a=136:
After finishing the pre-service that original face image is carried out, enter step 102.In step 102, calculate integrogram through pretreated face image.The common practise that the calculating of the integrogram of face image is belonged to this area, do not repeat them here, specifically can be referring to document 1 " Paul Viola; Micheal Jones; Rapid Object Detection using a Boosted Cascade of Simple Features " and document 2 " Yubo WANG, Haizhou Al, Bo WU; Chang HUANG, Real TimeFacial Expression Recognition with Adaboost ".
Then, in step 103, with the counting value returns of Weak Classifier.Wherein, Weak Classifier h (x, f, p, θ) by a feature f, the p of threshold value θ and indication sign of inequality direction forms:
Wherein, f (x) is the eigenwert of Weak Classifier, and f (x) gets the haar eigenwert, so the eigenwert of Weak Classifier is exactly the haar eigenwert, and equally can be about the detailed calculated method of haar feature referring to above-mentioned document 1 and document 2.θ is used to judge whether the eigenwert of Weak Classifier satisfies the predetermined threshold value of smiling face's condition, and its value can obtain by a large amount of training sample training.P is used to adjust the direction of the sign of inequality, as f (x) during, p=1 less than threshold value θ, and as f (x) during greater than threshold value θ, p=-1; The ballot of representing Weak Classifier during h=1 is the smiling face, and the ballot of representing Weak Classifier during h=0 is non-smiling face.Usually, the Weak Classifier of participation smiling face identification has a plurality of, for example 300.
Then, in step 104, whether the count value of judging Weak Classifier less than the sum of Weak Classifier, if less than the sum of Weak Classifier, then enter step 105; If more than or equal to the sum of Weak Classifier, then enter step 108;
In step 105, calculate the eigenwert of current Weak Classifier, and whether the eigenwert of Weak Classifier of judging current calculating is less than the threshold value θ of current Weak Classifier, wherein, the count value of Weak Classifier is the current Weak Classifier index value at all Weak Classifiers, during less than threshold value θ, enters step 106 in the eigenwert of the Weak Classifier of judging current calculating, during more than or equal to threshold value θ, enter step 107 in the eigenwert of the Weak Classifier of judging current calculating.
In step 106, the alpha value of the current Weak Classifier that adds up.Wherein, the alpha value is the weighted value of corresponding Weak Classifier, and its value can obtain by the training sample training.
In step 107, the count value of Weak Classifier is added up.And after step 107, get back to step 104.
When the count value of Weak Classifier during, enter step 108 more than or equal to Weak Classifier total.In step 108, judge that whether the weighted value that adds up is greater than predetermined threshold value n.If the accumulative total of alpha value and greater than threshold value n then enters step 109, if the accumulative total of alpha value and be less than or equal to threshold value n then enters step 110.In the present embodiment, threshold value n is half of summation of the weighted value (alpha value) of all Weak Classifiers.
In step 109, calculate smiling face's scoring, the result that the smiling face detects is smiling face's scoring of calculating, and output smiling face scoring.
In step 110, judge that result that the smiling face detects is not for being the smiling face.
In step 111, the result that the output smiling face detects.Such as, if enter into this step by step 109, the then smiling face scoring of the result of the smiling face of output detection for calculating; If enter into this step by step 110, then the result that detects of the smiling face of output is not for being the smiling face.
Be not difficult to find that step 102 in the present embodiment to step 110 detects for the smiling face that the Adaboost algorithm with voting mechanism carries out.Adaboost that computation complexity is low ballot principle is incorporated in smiling face's detection algorithm, can make the computation complexity of testing process reduce significantly, thereby can be poor in computing power, realize that real-time smiling face detects on the portable set a little less than the storage capacity.And, because it is very sensitive to noise based on the Adaboost algorithm of voting mechanism, to the difficult identification of smiling face and non-smiling face's a shade of difference, therefore before employing Adaboost algorithm carries out smiling face's detection, by denoising and brightness regulation, original face image has been carried out pre-service earlier, the insufficient section that Adaboost is applied to smiling face's identification improves, make and in reducing testing process, in the computation complexity, also can guarantee the accuracy that detects.
Second embodiment of the invention relates to a kind of smiling face detection method.Second embodiment and first embodiment are roughly the same, and its difference is:
In the first embodiment, be earlier original face image to be carried out denoising, again the face image after denoising is carried out brightness regulation and handle.And in the present embodiment, be to carry out brightness regulation earlier to handle, again the face image after handling through brightness regulation being carried out denoising, face image that will be after denoising carries out smiling face's detection with the Adaboost algorithm of voting mechanism.
Need to prove that each method embodiment of the present invention all can be realized in modes such as software, hardware, firmwares.No matter the present invention be with software, hardware, or the firmware mode realize, instruction code can be stored in the storer of computer-accessible of any kind (for example permanent or revisable, volatibility or non-volatile, solid-state or non-solid-state, fixing or removable medium or the like).Equally, storer can for example be programmable logic array (Programmable ArrayLogic, be called for short " PAL "), random access memory (Random Access Memory, be called for short " RAM "), programmable read only memory (Programmable Read Only Memory, be called for short " PROM "), ROM (read-only memory) (Read-Only Memory, be called for short " ROM "), Electrically Erasable Read Only Memory (Electrically Erasable Programmable ROM, be called for short " EEPROM "), disk, CD, digital versatile disc (Digital Versatile Disc is called for short " DVD ") or the like.
Third embodiment of the invention relates to a kind of smiling face's detection system.Fig. 4 is the structural representation of this smiling face's detection system.This smiling face's detection system comprises:
Pretreatment module is used for original face image is carried out pre-service, and pre-service comprises denoising and brightness regulation processing.
Detection module is used for carrying out smiling face's detection with through the pretreated face image of pretreatment module with the Adaboost algorithm.
Output module is used to export the result that smiling face that detection module carries out detects.
Wherein, pretreatment module comprises following submodule:
The denoising submodule is used for original face image is carried out denoising.
The brightness regulation processing sub is used for that the face image after handling through the denoising submodule is carried out brightness regulation and handles.
Specifically, the denoising submodule further comprises:
First calculating sub module is used for calculating the derivative of each pixel on line direction and column direction of face image, triggers second calculating sub module.
Second calculating sub module is used for according to the derivative of each pixel on line direction and column direction that calculates, and Grad and the gradient-norm of calculating each pixel are long, trigger the 3rd calculating sub module.
The 3rd calculating sub module is used for according to the Grad of each pixel and gradient-norm long, tries to achieve the divergence of the ratio of the Grad of each pixel and gradient-norm length, triggers the noise reduction process submodule.
The noise reduction process submodule is used for according to following formula face image u
iCarry out once will making an uproar processing, and trigger the iteration submodule:
Wherein, i is initialized as 0, u
0Represent original face image, u
I+1Expression is to u
iCarry out the face image after the noise reduction process one time,
Represent the divergence of the ratio of the Grad of each pixel and gradient-norm length, the elements of a fix of x and y remarked pixel, t are the algorithm step-length that is used to regulate algorithm speed, and λ is default adjustable parameter, and the more little expression of λ requires image smooth more.
The iteration submodule is used for judgement || u
I+1-u
i|| whether＜ε sets up, and wherein ε is given threshold value, if || u
I+1-u
i||＜ε sets up, then with u
I+1As the face image after denoising.If || u
I+1-u
i||＜ε is false, and then with i=i+1, triggers first calculating sub module again.
The brightness regulation processing sub further comprises:
The gray average calculating sub module is used for calculating the average of the gray scale of all pixels of face image to be regulated.
Adjust submodule, be used for the average of the gray scale of all pixels of calculating according to the gray average calculating sub module,, adjust to a definite value the gray scale of all pixels in the face image to be regulated.
In the Adaboost of present embodiment algorithm, the Weak Classifier that is used to vote be h (x, f, p, θ),
Wherein, f (x) is the eigenwert of Weak Classifier, and f (x) gets the haar eigenwert, and θ is used to judge whether the eigenwert of Weak Classifier satisfies the predetermined threshold value of smiling face's condition, threshold value θ can obtain by the training sample training, p is used to adjust the direction of the sign of inequality, as f (x) during less than threshold value θ, and p=1, as f (x) during greater than threshold value θ, p=-1 represents during h=1 that the ballot of Weak Classifier is the smiling face, and the ballot of representing Weak Classifier during h=0 is non-smiling face.
Specifically, detection module comprises following submodule:
The integrogram calculating sub module is used to calculate the integrogram through pretreated face image.
The submodule that makes zero is used for the counting value returns with Weak Classifier.
Count value is judged submodule, whether the count value that is used to judge Weak Classifier less than the sum of Weak Classifier, if less than the sum of Weak Classifier, then trigger the eigenvalue calculation submodule, if more than or equal to the sum of Weak Classifier, then trigger thresholding and judge submodule;
The eigenvalue calculation submodule is used to calculate the eigenwert of current Weak Classifier, and whether the eigenwert of Weak Classifier of judging current calculating is less than predetermined threshold value θ, wherein, the count value of Weak Classifier is the current Weak Classifier index value at all Weak Classifiers, the eigenvalue calculation submodule is in the eigenwert of the Weak Classifier of judging current calculating during less than threshold value θ, triggers the weighted value submodule that adds up; The eigenvalue calculation submodule is in the eigenwert of the Weak Classifier of judging current calculating during more than or equal to threshold value θ, the flip-flop number value submodule that adds up;
Weighted value submodule be used to the to add up weighted value of current Weak Classifier that adds up, and the flip-flop number value submodule that adds up;
The count value submodule that adds up is used for the count value of Weak Classifier is added up, and the flip-flop number value is judged submodule;
Thresholding judges that whether submodule is used to judge the weighted value that adds up weighted value submodule adds up greater than predetermined threshold value n, if greater than threshold value n, and the then smiling face scoring of result that detect of smiling face for calculating; If be less than or equal to threshold value n, then the result of smiling face's detection is non-smiling face's image.Wherein, threshold value n is half of summation of the weighted value of all Weak Classifiers.
Be not difficult to find that first embodiment is and the corresponding method embodiment of present embodiment, present embodiment can with the enforcement of working in coordination of first embodiment.The correlation technique details of mentioning in first embodiment is still effective in the present embodiment, in order to reduce repetition, repeats no more here.Correspondingly, the correlation technique details of mentioning in the present embodiment also can be applicable in first embodiment.
Four embodiment of the invention relates to a kind of smiling face detection method.The 4th embodiment and the 3rd embodiment are roughly the same, and its difference is:
In the 3rd embodiment, pretreatment module is earlier original face image to be carried out denoising, again the face image after denoising is carried out brightness regulation and handles.And in the present embodiment, pretreatment module is to carry out brightness regulation earlier to handle, and again the face image after handling through brightness regulation is carried out denoising.That is to say, by the brightness regulation processing sub original face image is carried out brightness regulation earlier and handle, by the denoising submodule face image after handling through the brightness regulation processing sub is carried out denoising again.
Be not difficult to find that second embodiment is and the corresponding method embodiment of present embodiment, present embodiment can with the enforcement of working in coordination of second embodiment.The correlation technique details of mentioning in second embodiment is still effective in the present embodiment, in order to reduce repetition, repeats no more here.Correspondingly, the correlation technique details of mentioning in the present embodiment also can be applicable in second embodiment.
Need to prove, each unit of mentioning in each equipment embodiment of the present invention all is a logical block, physically, a logical block can be a physical location, it also can be the part of a physical location, can also realize that the physics realization mode of these logical blocks itself is not most important with the combination of a plurality of physical locations, the combination of the function that these logical blocks realized is the key that just solves technical matters proposed by the invention.In addition, for outstanding innovation part of the present invention, above-mentioned each the equipment embodiment of the present invention will not introduced not too close unit with solving technical matters relation proposed by the invention, and this does not show that there is not other unit in the said equipment embodiment.
Though pass through with reference to some of the preferred embodiment of the invention, the present invention is illustrated and describes, but those of ordinary skill in the art should be understood that and can do various changes to it in the form and details, and without departing from the spirit and scope of the present invention.
Claims (15)
1. a smiling face detection method is characterized in that, comprises following steps:
Original face image is carried out pre-service, and described pre-service comprises denoising and brightness regulation processing;
Will be through described pretreated face image, carry out smiling face's detection with the Adaboost algorithm of voting mechanism;
Export the result that described smiling face detects.
2. smiling face detection method according to claim 1 is characterized in that, described original face image is gone in the pretreated step, carries out described denoising earlier, carries out described brightness regulation again and handles.
3. smiling face detection method according to claim 1 is characterized in that, described denoising is the face image denoising based on the variational method, comprises following substep:
The derivative of each pixel on line direction and column direction in A1, the calculating face image;
A2, according to the derivative of each pixel on line direction and column direction that calculates, Grad and the gradient-norm of calculating each pixel are long;
A3, long according to the Grad and the gradient-norm of each pixel tries to achieve the divergence of the ratio of the Grad of each pixel and gradient-norm length;
A4, according to following formula to face image u
iCarry out noise reduction process one time:
Wherein, i is initialized as 0, u
0Represent original face image, u
I+1Expression is to u
iCarry out the face image after the noise reduction process one time,
Represent the divergence of the ratio of the Grad of each pixel and gradient-norm length, the elements of a fix of x and y remarked pixel, t are the algorithm step-length that is used to regulate algorithm speed, and λ is default adjustable parameter, and the more little expression of λ requires image smooth more;
A5, judgement || u
I+1-u
i|| whether＜ε sets up, and wherein said ε is given threshold value, if || u
I+1-u
i||＜ε sets up, then with u
I+1As the face image after described denoising; If || u
I+1-u
i||＜ε is false, and then with i=i+1, repeats described steps A 1 to described steps A 5.
4. smiling face detection method according to claim 3 is characterized in that, described t and described λ are 0.01.
5. smiling face detection method according to claim 1 is characterized in that, described brightness regulation is handled and comprised following substep:
The average of the gray scale of all pixels in the calculating face image to be regulated;
According to the average of the gray scale of all pixels of described calculating,, adjust to a definite value with the gray scale of all pixels in the face image described to be regulated.
6. according to each described smiling face detection method in the claim 1 to 5, it is characterized in that, in described Adaboost algorithm, the Weak Classifier that is used to vote be h (x, f, p, θ),
Wherein, f (x) is the eigenwert of described Weak Classifier, described f (x) gets the haar eigenwert, θ is used to judge whether the eigenwert of described Weak Classifier satisfies the predetermined threshold value of smiling face's condition, p is used to adjust the direction of the sign of inequality, as described f (x) during less than described threshold value θ, and described p=1, as described f (x) during greater than described threshold value θ, described p=-1; The ballot of representing Weak Classifier during h=1 is the smiling face, and the ballot of representing Weak Classifier during h=0 is non-smiling face.
7. smiling face detection method according to claim 6 is characterized in that, describedly carries out comprising following substep in the step that the smiling face detects with the Adaboost algorithm:
B1, calculate integrogram through described pretreated face image;
B2, with the counting value returns of Weak Classifier;
B3, judge Weak Classifier count value whether less than the sum of Weak Classifier, if, then enter step B4 less than the sum of Weak Classifier; If more than or equal to the sum of Weak Classifier, then enter step B7;
The eigenwert of B4, the current Weak Classifier of calculating, and whether the eigenwert of Weak Classifier of judging current calculating is less than described threshold value θ, wherein, the count value of Weak Classifier is the current Weak Classifier index value at all Weak Classifiers, during less than described threshold value θ, enter step B5 in the eigenwert of the Weak Classifier of judging current calculating; During more than or equal to described threshold value θ, enter step B6 in the eigenwert of the Weak Classifier of judging current calculating;
The weighted value of B5, the current Weak Classifier that adds up;
B6, the count value of Weak Classifier is added up, get back to described step B3;
Whether the weighted value that B7, judgement add up is greater than predetermined threshold value n, if greater than described threshold value n, the result that then described smiling face detects marks for the smiling face who calculates; If be less than or equal to described threshold value n, the result that then described smiling face detects is non-smiling face's image.
8. smiling face detection method according to claim 7 is characterized in that, described threshold value θ obtains by training sample training, and described threshold value n is half of summation of the weighted value of all Weak Classifiers.
9. smiling face's detection system is characterized in that, comprises:
Pretreatment module is used for original face image is carried out pre-service, and described pre-service comprises denoising and brightness regulation processing;
Detection module is used for through the pretreated face image of described pretreatment module, carries out smiling face's detection with the Adaboost algorithm of voting mechanism;
Output module is used to export the result that smiling face that described detection module carries out detects.
10. smiling face's detection system according to claim 9 is characterized in that, described pretreatment module comprises following submodule:
The denoising submodule is used for described original face image is carried out denoising;
The brightness regulation processing sub is used for that the face image after handling through described denoising submodule is carried out brightness regulation and handles.
11. smiling face's detection system according to claim 9 is characterized in that, described denoising submodule comprises:
First calculating sub module is used for calculating the derivative of each pixel on line direction and column direction of face image, triggers second calculating sub module;
Described second calculating sub module is used for according to the derivative of each pixel on line direction and column direction that calculates, and Grad and the gradient-norm of calculating each pixel are long, trigger the 3rd calculating sub module;
Described the 3rd calculating sub module is used for according to the Grad and the gradient-norm of each pixel long, tries to achieve the divergence of the ratio of the Grad of each pixel and gradient-norm length, triggers the noise reduction process submodule;
Described noise reduction process submodule is used for according to following formula face image u
iCarry out once will making an uproar processing, and trigger the iteration submodule:
Wherein, i is initialized as 0, u
0Represent original face image, u
I+1Expression is to u
iCarry out the face image after the noise reduction process one time,
Represent the divergence of the ratio of the Grad of each pixel and gradient-norm length, the elements of a fix of x and y remarked pixel, t are the algorithm step-length that is used to regulate algorithm speed, and λ is default adjustable parameter, and the more little expression of λ requires image smooth more;
Described iteration submodule is used for judgement || u
I+1-u
i|| whether＜ε sets up, and wherein said ε is given threshold value, if || u
I+1-u
i||＜ε sets up, then with u
I+1As the face image after described denoising; If || u
I+1-u
i||＜ε is false, and then with i=i+1, triggers described first calculating sub module again.
12. smiling face's detection system according to claim 9 is characterized in that, described brightness regulation processing sub comprises:
The gray average calculating sub module is used for calculating the average of the gray scale of all pixels of face image to be regulated;
Adjust submodule, be used for the average of the gray scale of all pixels of calculating according to described gray average calculating sub module,, adjust to a definite value the gray scale of all pixels in the face image described to be regulated.
13. according to each described smiling face's detection system in the claim 9 to 12, it is characterized in that, in described Adaboost algorithm, the Weak Classifier that is used to vote be h (x, f, p, θ),
Wherein, f (x) is the eigenwert of described Weak Classifier, described f (x) gets the haar eigenwert, θ is used to judge whether the eigenwert of described Weak Classifier satisfies the predetermined threshold value of smiling face's condition, p is used to adjust the direction of the sign of inequality, as described f (x) during less than described threshold value θ, and described p=1, as described f (x) during greater than described threshold value θ, described p=-1; The ballot of representing Weak Classifier during h=1 is the smiling face, and the ballot of representing Weak Classifier during h=0 is non-smiling face.
14. smiling face's detection system according to claim 13 is characterized in that, described detection module comprises following submodule:
The integrogram calculating sub module is used to calculate the integrogram through described pretreated face image;
The submodule that makes zero is used for the counting value returns with Weak Classifier;
Count value is judged submodule, whether the count value that is used to judge Weak Classifier less than the sum of Weak Classifier, if less than the sum of Weak Classifier, then trigger the eigenvalue calculation submodule, if more than or equal to the sum of Weak Classifier, then trigger thresholding and judge submodule;
Described eigenvalue calculation submodule is used to calculate the eigenwert of current Weak Classifier, and whether the eigenwert of Weak Classifier of judging current calculating is less than described threshold value θ, wherein, the count value of Weak Classifier is the current Weak Classifier index value at all Weak Classifiers, described eigenvalue calculation submodule is in the eigenwert of the Weak Classifier of judging current calculating during less than described threshold value θ, triggers the weighted value submodule that adds up; Described eigenvalue calculation submodule is in the eigenwert of the Weak Classifier of judging current calculating during more than or equal to described threshold value θ, the flip-flop number value submodule that adds up;
The described weighted value submodule that adds up is used to the weighted value of current Weak Classifier that adds up, and triggers the described count value submodule that adds up;
The described count value submodule that adds up is used for the count value of Weak Classifier is added up, and triggers described count value and judge submodule;
Described thresholding judges that whether submodule is used to judge the weighted value that adds up described weighted value submodule adds up greater than predetermined threshold value n, if greater than described threshold value n, and the smiling face scoring of the result that then described smiling face detects for calculating; If be less than or equal to described threshold value n, the result that then described smiling face detects is non-smiling face's image.
15. smiling face's detection system according to claim 14 is characterized in that, described threshold value θ obtains by training sample training, and described threshold value n is half of summation of the weighted value of all Weak Classifiers.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201010276313 CN101950356B (en) | 2010-09-09 | 2010-09-09 | Smiling face detecting method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201010276313 CN101950356B (en) | 2010-09-09 | 2010-09-09 | Smiling face detecting method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101950356A true CN101950356A (en) | 2011-01-19 |
| CN101950356B CN101950356B (en) | 2013-08-28 |
Family
ID=43453851
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201010276313 Active CN101950356B (en) | 2010-09-09 | 2010-09-09 | Smiling face detecting method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101950356B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012139271A1 (en) * | 2011-04-11 | 2012-10-18 | Intel Corporation | Smile detection techniques |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080025576A1 (en) * | 2006-07-25 | 2008-01-31 | Arcsoft, Inc. | Method for detecting facial expressions of a portrait photo by an image capturing electronic device |
| CN101699470A (en) * | 2009-10-30 | 2010-04-28 | 华南理工大学 | Extracting method for smiling face identification on picture of human face |
| CN101702199A (en) * | 2009-11-13 | 2010-05-05 | 深圳华为通信技术有限公司 | Smiling face detection method and device and mobile terminal |
-
2010
- 2010-09-09 CN CN 201010276313 patent/CN101950356B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080025576A1 (en) * | 2006-07-25 | 2008-01-31 | Arcsoft, Inc. | Method for detecting facial expressions of a portrait photo by an image capturing electronic device |
| CN101699470A (en) * | 2009-10-30 | 2010-04-28 | 华南理工大学 | Extracting method for smiling face identification on picture of human face |
| CN101702199A (en) * | 2009-11-13 | 2010-05-05 | 深圳华为通信技术有限公司 | Smiling face detection method and device and mobile terminal |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012139271A1 (en) * | 2011-04-11 | 2012-10-18 | Intel Corporation | Smile detection techniques |
| US9268995B2 (en) | 2011-04-11 | 2016-02-23 | Intel Corporation | Smile detection techniques |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101950356B (en) | 2013-08-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10635890B2 (en) | Facial recognition method and apparatus, electronic device, and storage medium | |
| US20200160040A1 (en) | Three-dimensional living-body face detection method, face authentication recognition method, and apparatuses | |
| US10523894B2 (en) | Automated selection of keeper images from a burst photo captured set | |
| US7376270B2 (en) | Detecting human faces and detecting red eyes | |
| Theocharides et al. | A parallel architecture for hardware face detection | |
| US8085996B2 (en) | Image processing apparatus, image display apparatus, imaging apparatus, method for image processing therefor, and program | |
| US8345921B1 (en) | Object detection with false positive filtering | |
| US9836640B2 (en) | Face detector training method, face detection method, and apparatuses | |
| US20110194779A1 (en) | Apparatus and method for detecting multi-view specific object | |
| US20060222243A1 (en) | Extraction and scaled display of objects in an image | |
| US20230030267A1 (en) | Method and apparatus for selecting face image, device, and storage medium | |
| CN108921131B (en) | Method and device for generating face detection model and three-dimensional face image | |
| CN103617432A (en) | Method and device for recognizing scenes | |
| US8421877B2 (en) | Feature data compression device, multi-directional face detection system and detection method thereof | |
| EP2370931B1 (en) | Method, apparatus and computer program product for providing an orientation independent face detector | |
| CN101339609A (en) | Image processing apparatus and image processing method | |
| US20140133758A1 (en) | Real-time face detection using pixel pairs | |
| CN111696080B (en) | Face fraud detection method, system and storage medium based on static texture | |
| CN111325237B (en) | Image recognition method based on attention interaction mechanism | |
| US8520955B2 (en) | Object detection apparatus and method | |
| Ren et al. | Real-time optimization of Viola-Jones face detection for mobile platforms | |
| Pettersson et al. | The histogram feature-a resource-efficient weak classifier | |
| Andiani et al. | Face recognition for work attendance using multitask convolutional neural network (MTCNN) and pre-trained facenet | |
| CN101276404A (en) | System and method for quickly and exactly processing intelligent image | |
| CN101950356B (en) | Smiling face detecting method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20170210 Address after: Room 32, building 3205F, No. 707, Zhang Yang Road, free trade zone,, China (Shanghai) Patentee after: Xin Xin Finance Leasing Co.,Ltd. Address before: 201203 Shanghai city Zuchongzhi road Pudong New Area Zhangjiang hi tech park, Spreadtrum Center Building 1, Lane 2288 Patentee before: SPREADTRUM COMMUNICATIONS (SHANGHAI) Co.,Ltd. |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20170712 Address after: 100033 room 2062, Wenstin Executive Apartment, 9 Financial Street, Beijing, Xicheng District Patentee after: Xin Xin finance leasing (Beijing) Co.,Ltd. Address before: Room 32, building 707, Zhang Yang Road, China (Shanghai) free trade zone, 3205F Patentee before: Xin Xin Finance Leasing Co.,Ltd. |
|
| TR01 | Transfer of patent right | ||
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20110119 Assignee: SPREADTRUM COMMUNICATIONS (SHANGHAI) Co.,Ltd. Assignor: Xin Xin finance leasing (Beijing) Co.,Ltd. Contract record no.: 2018990000163 Denomination of invention: Smiling face detecting method and system Granted publication date: 20130828 License type: Exclusive License Record date: 20180626 |
|
| TR01 | Transfer of patent right |
Effective date of registration: 20200306 Address after: 201203 Zuchongzhi Road, China (Shanghai) pilot Free Trade Zone, Pudong New Area, Shanghai 2288 Patentee after: SPREADTRUM COMMUNICATIONS (SHANGHAI) Co.,Ltd. Address before: 100033 room 2062, Wenstin administrative apartments, 9 Financial Street B, Xicheng District, Beijing. Patentee before: Xin Xin finance leasing (Beijing) Co.,Ltd. |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20200603 Address after: 361012 unit 05, 8th floor, building D, Xiamen international shipping center, No.97 Xiangyu Road, Xiamen area, Xiamen pilot Free Trade Zone, Xiamen City, Fujian Province Patentee after: Xinxin Finance Leasing (Xiamen) Co.,Ltd. Address before: 201203 Zuchongzhi Road, China (Shanghai) pilot Free Trade Zone, Pudong New Area, Shanghai 2288 Patentee before: SPREADTRUM COMMUNICATIONS (SHANGHAI) Co.,Ltd. |
|
| TR01 | Transfer of patent right | ||
| EC01 | Cancellation of recordation of patent licensing contract |
Assignee: SPREADTRUM COMMUNICATIONS (SHANGHAI) Co.,Ltd. Assignor: Xin Xin finance leasing (Beijing) Co.,Ltd. Contract record no.: 2018990000163 Date of cancellation: 20210301 |
|
| EC01 | Cancellation of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20110119 Assignee: SPREADTRUM COMMUNICATIONS (SHANGHAI) Co.,Ltd. Assignor: Xinxin Finance Leasing (Xiamen) Co.,Ltd. Contract record no.: X2021110000010 Denomination of invention: Smiling face detection method and system Granted publication date: 20130828 License type: Exclusive License Record date: 20210317 |
|
| EE01 | Entry into force of recordation of patent licensing contract | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20230714 Address after: 201203 Shanghai city Zuchongzhi road Pudong New Area Zhangjiang hi tech park, Spreadtrum Center Building 1, Lane 2288 Patentee after: SPREADTRUM COMMUNICATIONS (SHANGHAI) Co.,Ltd. Address before: 361012 unit 05, 8 / F, building D, Xiamen international shipping center, 97 Xiangyu Road, Xiamen area, Xiamen pilot Free Trade Zone, Fujian Province Patentee before: Xinxin Finance Leasing (Xiamen) Co.,Ltd. |
|
| TR01 | Transfer of patent right |