CN115776410B - Face data encryption transmission method for terminal identity authentication - Google Patents
Face data encryption transmission method for terminal identity authentication Download PDFInfo
- Publication number
- CN115776410B CN115776410B CN202310043171.1A CN202310043171A CN115776410B CN 115776410 B CN115776410 B CN 115776410B CN 202310043171 A CN202310043171 A CN 202310043171A CN 115776410 B CN115776410 B CN 115776410B
- Authority
- CN
- China
- Prior art keywords
- hog
- image
- gradient
- hog characteristic
- characteristic 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.)
- Active
Links
Images
Abstract
The invention relates to the technical field of data encryption, in particular to a face data encryption transmission method for terminal identity authentication, which comprises the following steps: the method comprises the steps of collecting face images, dividing the face images to obtain a plurality of identical image blocks, obtaining first importance degrees of HOG feature images, obtaining second importance degrees of HOG feature images, obtaining comprehensive importance degrees of each HOG feature image, obtaining final HOG feature images of each image block, obtaining encrypted images, and transmitting.
Description
Technical Field
The invention relates to the technical field of data encryption, in particular to a face data encryption transmission method for terminal identity authentication.
Background
Face recognition is a biological recognition technology for carrying out identity recognition based on facial feature information of people, and implementation scenes of face authentication for terminals are also growing day by day, such as face recognition of mobile phones during payment and places with scene restrictions in and out for high-speed rail stations and the like, so that encryption of face data is synchronously carried out in order to prevent face data from being cracked, so that payment loss is caused.
The traditional encryption mode is to normalize the gradient direction histogram corresponding to the HOG characteristic image to achieve the attack mode of resisting statistical analysis, but the conversion mode is single, other attack modes cannot be effectively resisted, and especially when violent cracking is encountered, the safety of data is difficult to ensure.
Disclosure of Invention
The invention provides a face data encryption transmission method for terminal identity authentication, which aims to solve the problem that the safety of data is difficult to ensure when the existing face data encounters violent cracking.
The invention discloses a face data encryption transmission method for terminal identity authentication, which adopts the following technical scheme:
acquiring a face image, and dividing the face image to obtain a plurality of identical image blocks;
acquiring HOG characteristic images of each image block, and taking standard deviations of gradient amplitudes of all gradient directions in each HOG characteristic image as a first importance degree of each HOG characteristic image;
acquiring the overall gradient direction of each HOG characteristic image according to the horizontal gradient amplitude component and the vertical gradient amplitude component of the gradient amplitudes of all gradient directions in each HOG characteristic image; acquiring a second important degree of each HOG characteristic image according to the direction deviation value of the overall gradient direction of each HOG characteristic image and the overall gradient direction of each HOG characteristic image in the neighborhood and the gradient amplitude value difference value of each HOG characteristic image and each HOG characteristic image in the neighborhood in the corresponding gradient direction;
Acquiring the comprehensive importance degree of each HOG characteristic image according to the first importance degree and the second importance degree;
acquiring initial transformation gradient amplitude values of each gradient direction according to the gradient amplitude value average value of each gradient direction and the gradient amplitude value of each gradient direction in each HOG characteristic image; obtaining a target gradient amplitude value of each gradient direction in each HOG characteristic image by utilizing the gradient amplitude value of each gradient direction in each HOG characteristic image and the initial transformation gradient amplitude value of each gradient direction;
obtaining a transformation coefficient of each gradient direction according to the initial transformation gradient amplitude of each gradient direction and the target gradient amplitude, obtaining a final transformation gradient amplitude of each gradient direction of the HOG characteristic image according to the transformation coefficient of each gradient direction, the initial transformation gradient amplitude and the comprehensive importance degree of the HOG characteristic image, and obtaining a final HOG characteristic image of each image block according to all final transformation gradient amplitudes;
and obtaining an encrypted image according to all the final HOG characteristic images, and transmitting the encrypted image, the transformation coefficient corresponding to each final HOG characteristic image and the comprehensive importance degree.
Preferably, the method further comprises:
acquiring each HOG characteristic image and a target HOG characteristic image symmetrical about the vertical center line of the face image;
According to the overall gradient direction of each HOG characteristic image and the corresponding target HOG characteristic image, obtaining the direction symmetry degree of each HOG characteristic image in the neighborhood of the HOG characteristic image and the HOG characteristic image at the symmetrical position in the neighborhood of the corresponding target HOG characteristic image;
according to the direction deviation value of the overall gradient direction of each HOG characteristic image in the neighborhood of the HOG characteristic image and the HOG characteristic image in the symmetrical position in the neighborhood of each HOG characteristic image and the corresponding target HOG characteristic image, acquiring the overall gradient direction symmetry degree of the HOG characteristic image and the corresponding target HOG characteristic image in the neighborhood range;
judging whether to carry out neighborhood size adjustment according to the sum value of the direction deviation values of the whole gradient directions of the HOG characteristic images and all HOG characteristic images in the neighborhood of the HOG characteristic images, and obtaining the final neighborhood size corresponding to each HOG characteristic image;
normalizing according to the length of the final neighborhood size corresponding to each HOG characteristic image to obtain the necessity degree of the HOG characteristic image;
acquiring an adjustment value of each HOG characteristic image according to the corresponding necessity degree of the HOG characteristic image, the symmetry degree of the whole gradient direction and the sum of the gradient magnitudes of the HOG characteristic image and the corresponding target HOG characteristic image;
And adjusting the comprehensive importance degree of each corresponding HOG characteristic image according to the adjustment value to obtain the final comprehensive importance degree.
Preferably, obtaining the final neighborhood size includes:
when the sum of the direction deviation values of the whole gradient directions of the HOG characteristic images and all HOG characteristic images in the neighborhood of the HOG characteristic images is smaller than a preset sum threshold, gradually increasing the neighborhood size of the HOG characteristic images to obtain an increased target neighborhood;
and acquiring the sum value of the direction deviation values of the overall gradient directions of the HOG characteristic images and each HOG characteristic image in the target neighborhood, and taking the size of the corresponding target neighborhood as the final neighborhood size when the sum value of the direction deviation values of the overall gradient directions of the HOG characteristic images and each HOG characteristic image in the target neighborhood is larger than or equal to a preset sum value threshold.
Preferably, the obtaining the necessity of the HOG feature image includes:
acquiring the length of the corresponding final neighborhood size and the length of the minimum neighborhood size in the lengths of the final neighborhood sizes corresponding to all HOG characteristic images;
obtaining a target neighborhood size corresponding to the normalized HOG feature image according to the length of the corresponding final neighborhood size, the length of the minimum neighborhood size and the length of the final neighborhood size corresponding to each HOG feature image;
And taking the length corresponding to the target neighborhood size as the necessity of the HOG characteristic image.
Preferably, a calculation formula of a degree of directional symmetry of each HOG feature image in the neighborhood of the HOG feature image and the HOG feature image at a symmetrical position in the neighborhood of the corresponding target HOG feature image:
in the method, in the process of the invention,
representing HOG feature images
HOG feature images in the neighborhood of (2)
And corresponding target HOG characteristic image
In the neighborhood of (a) with HOG feature images
The degree of directional symmetry of the position-symmetric HOG feature image;
Preferably, the obtaining the overall gradient direction symmetry degree of the HOG feature image and the corresponding target HOG feature image in the neighborhood range includes:
acquiring a direction deviation value of the overall gradient direction of each HOG characteristic image in the neighborhood of the HOG characteristic image and the sum value of the ratio of the direction symmetry degree of each HOG characteristic image in the neighborhood of the HOG characteristic image and the HOG characteristic image in the symmetrical position in the neighborhood of the corresponding target HOG characteristic image;
And taking the sum value as the overall gradient direction symmetry degree of the HOG characteristic image and the corresponding target HOG characteristic image in the neighborhood range.
Preferably, the calculation formula of the adjustment value of each HOG feature image is:
in the method, in the process of the invention,
representing HOG feature images
Is a set of the adjustment values of (2);
Preferably, acquiring the second importance of each HOG feature image includes:
acquiring the sum of absolute values of gradient amplitude difference values of the HOG characteristic image and the HOG characteristic image in the adjacent region in all corresponding gradient directions;
and taking the sum of products of the sum of absolute values of gradient amplitude differences corresponding to all HOG feature images in the neighborhood of the HOG feature images and the direction deviation value of the whole gradient direction as a second important degree of the HOG feature images.
Preferably, obtaining the final transformed gradient magnitude for each gradient direction of the HOG feature image comprises:
obtaining the product of the transformation coefficient and the comprehensive importance degree;
And obtaining the ratio of the initial transformation gradient amplitude corresponding to each gradient direction in the HOG characteristic image to the product of the transformation coefficient and the comprehensive importance degree, and taking the ratio as the final transformation gradient amplitude of each gradient direction of the HOG characteristic image.
Preferably, the method further comprises: decrypting the encrypted data, the decrypting step comprising:
acquiring the gradient before transformation of each gradient direction of the HOG characteristic image according to each final HOG characteristic image, the corresponding transformation coefficient and the comprehensive importance degree;
acquiring HOG characteristic images according to gradients before transformation of each gradient direction of the HOG characteristic images;
and obtaining a face image according to all the HOG characteristic images.
The face data encryption transmission method for terminal identity authentication has the beneficial effects that:
because the gradient directions of the areas corresponding to different facial organs in the face image are different, the face image is divided into a plurality of image blocks, then the HOG characteristic image representing each image block is obtained, the first importance degree of the HOG characteristic image is reflected by the standard deviation of the gradient amplitude of the HOG characteristic image corresponding to all gradient directions, the larger the standard deviation is, the larger the gradient amplitude difference is, the corresponding first importance degree is also larger, namely the first importance degree reflects the importance degree of the HOG characteristic image, then the second importance degree of the HOG characteristic image is reflected by the direction deviation values and the gradient amplitude difference values of the HOG characteristic image and all HOG characteristic images in the neighborhood of the HOG characteristic image, namely the second importance degree reflects the importance degree of the HOG characteristic image and the HOG characteristic of the HOG characteristic image in the domain of the HOG characteristic image, so that the first importance degree of the HOG characteristic image is combined, comprehensively evaluating the comprehensive importance degree of the HOG characteristic images with the second importance degree of the HOG characteristic images in the HOG characteristic images and the HOG characteristic images in the HOG characteristic image field so that when the HOG characteristic images are attacked, the cracking result of all the HOG characteristic images after the HOG characteristic images are in error, therefore, the HOG characteristic images have avalanche effect during cracking, so that the safety of encrypted data obtained by encryption based on the comprehensive importance degree is higher, secondly, the comprehensive importance degree characterizes the significance of the gradient directions in the HOG characteristic images, and since the direction gradient of the region corresponding to a human face organ is different from the forehead and cheek region, the significant gradient directions exist in the HOG characteristic images corresponding to the human face organ region, and if the magnitude of the significant gradient directions is transformed, the gradient magnitude of each HOG characteristic image in the whole image is similar, the difference between all HOG characteristic images of the whole image is small, so that even if stored encrypted data is stolen, the stored encrypted data cannot be cracked, gradients of the HOG characteristic images are transformed according to the comprehensive importance degree and transformation coefficients of each gradient direction to obtain final transformation gradient amplitude values, the final HOG characteristic images of each image block are obtained according to the final transformation gradient amplitude values, the gradient amplitude values corresponding to all the final HOG characteristic images are similar, namely, the comprehensive importance degree of the HOG characteristic images is utilized to be different, encryption of different HOG characteristic images is realized, the safety of transmitted data is further improved, meanwhile, each final HOG characteristic image corresponds to a secret key, the complexity of the secret key is increased, and the safety of the transmitted data is ensured.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of an embodiment of a face data encryption transmission method for terminal identity authentication according to the present invention;
fig. 2 is a schematic diagram of a HOG feature image and a symmetrical HOG feature image in an embodiment of a face data encryption transmission method for terminal identity authentication according to the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
An embodiment of a face data encryption transmission method for terminal identity authentication of the present invention, as shown in fig. 1, includes:
s1, acquiring a face image, and dividing the face image to obtain a plurality of identical image blocks;
specifically, the height of the collecting device is consistent with the height of the head of the collected person, and the collecting device faces the person to obtain the face image of the front face of the person, and the collected front face image is subjected to face region segmentation by semantic segmentation to divide the image into the size of
Wherein the segmentation block size is
Wherein
For the size of the face image acquired,
an empirical value of 50 is given (a particular size implementer may adjust to a particular scenario,
too large increases sharpness but the recognition speed decreases).
Thus, the face image is acquired and segmented to obtain a plurality of identical image blocks.
S2, acquiring a first importance degree of the HOG characteristic image;
because the gradient directions of the pixels corresponding to the face and the forehead area are unified, and the gradient directions of the pixels corresponding to the five sense organs are disordered, the HOG characteristic images corresponding to the five sense organs area are obviously different from the cheeks, namely, the HOG characteristic images corresponding to the forehead area have more obvious gradient directions, namely, have stronger identification property, therefore, the implementation needs to change the gradient amplitude values with the identification gradient directions, so that the gradient amplitude values in each HOG characteristic image in the whole face image are similar, namely, the gradient amplitude value between every two HOG characteristic images of the whole face image is smaller, and therefore, even if the subsequently stored data are stolen, the data are not easy to crack.
Therefore, in this embodiment, the HOG feature image of each image block is first obtained, and after each HOG feature image is obtained, each HOG feature image corresponds to a direction gradient histogram, so that standard deviation of gradient magnitudes in all gradient directions in the direction gradient histogram corresponding to the HOG feature image is obtained, and the standard deviation is used as a first importance degree of the HOG feature image, where a calculation formula of the first importance degree is as follows:
in the method, in the process of the invention,
representing HOG feature images
Standard deviation of gradient magnitudes in all gradient directions, i.e., a first importance level of the HOG feature image;
it should be noted that, the upper label 9 of the summation formula represents 9 angles obtained by equally dividing 180 degrees, namely 9 gradient directions corresponding to the direction gradient histogram, and the standard deviation represents the HOG characteristic image
The larger the standard deviation is, the larger the corresponding gradient amplitude difference is, namely, the gradient amplitude in one or more gradient directions is far larger than the gradient amplitudes in the other gradient directions, and secondly, the standard deviation formula is the formula of the prior art, and the embodiment is not repeated.
S3, acquiring a second important degree of the HOG characteristic image;
specifically, according to the horizontal gradient amplitude components and the vertical gradient amplitude components of the gradient amplitudes of all gradient directions in each HOG characteristic image, the whole gradient direction of each HOG characteristic image is obtained; and obtaining a second important degree of each HOG characteristic image according to the direction deviation value of the whole gradient direction of each HOG characteristic image and the whole gradient direction of each HOG characteristic image in the neighborhood and the gradient amplitude value difference value of each HOG characteristic image and each HOG characteristic image in the neighborhood in the corresponding gradient direction.
Specifically, a second important degree of each HOG feature image is acquired: calculating gradient direction of each pixel point in image block through HOG direction characteristics
And gradient magnitude
And performing statistics and putting into the obtained direction gradient histogram, wherein the abscissa of the direction gradient histogram is obtained by equally dividing 180 degrees into 9 angles, namely 9 gradient directions are obtained, and the 9 gradient directions are sequentially as follows:
the method comprises the steps of carrying out a first treatment on the surface of the During calculation, judging gradient direction sections of the pixel points in the gradient direction, wherein the gradient direction sections are sequentially as follows
，
……
And is assigned according to the maximum gradient direction, the minimum gradient direction and each gradient direction in each gradient direction interval, for example, when a pixel point in an image block 
Is the gradient direction of (a)
It exists in the gradient direction interval
In (3), it is allocated in the gradient direction interval
The maximum gradient direction of (2) is
The gradient amplitude assigned to the minimum gradient direction is
Wherein
Is a pixel point
The method for obtaining the horizontal gradient amplitude component and the vertical gradient amplitude component of the gradient amplitude in all gradient directions in the HOG feature image by using a triangle formula is not repeated in the embodiment, so that the horizontal gradient amplitude component and the vertical gradient amplitude component of the gradient amplitude in all gradient directions in the HOG feature image are obtained, the overall gradient direction of each HOG feature image is obtained, and the calculation formula of the overall gradient direction is as follows:
in the method, in the process of the invention,
representing HOG feature images
Is a global gradient direction of (2);
it should be noted that, in the formula, the superscript 9 of the summation formula is the total number of gradient directions of the HOG feature image, and the horizontal gradient amplitude component and the vertical gradient amplitude component in the formula are positive and negative based on the pointing direction, for example, when the gradient direction is 40 °, the corresponding gradient is in the horizontal direction 
Direction) of the gradient magnitude component is positive, and when the gradient direction is 150 DEG, the corresponding gradient is in the horizontal direction [ ]
Direction) is negative, and secondly, the calculation formula of the overall gradient direction is the formula of the prior art, which is not described in detail in this embodiment.
And (3) obtaining the overall gradient direction of each HOG characteristic image in the 8 neighborhood range of each HOG characteristic image by using a calculation formula of the overall gradient direction, and based on the overall gradient direction, obtaining a direction deviation value of the overall gradient direction corresponding to each HOG characteristic image in the vicinity of each HOG characteristic image and each HOG characteristic image in the vicinity of each HOG characteristic image, wherein the calculation formula of the direction deviation value of the overall gradient direction corresponding to each HOG characteristic image in the vicinity of each HOG characteristic image is as follows:
in the method, in the process of the invention,
representing HOG specialSign image
HOG feature images in its neighborhood
Direction deviation values of the overall gradient direction of (a);
note that, when the HOG feature image
Is characterized by the overall gradient direction of (a) and the HOG characteristic image in the adjacent region 
The greater the overall gradient directional deviation, i.e. the directional difference
The closer to 90 DEG, the larger the direction deviation value is, thereby representing the HOG feature image
In HOG feature images
The more obvious the identification in the local area is, wherein the denominator +1 is used for preventing the denominator from being 0, so that each part is obtainedAnd the direction deviation value of the overall gradient direction corresponding to each HOG characteristic image in the neighborhood of each HOG characteristic image.
Based on a calculation formula of a direction deviation value of an overall gradient direction corresponding to each HOG feature image in the HOG feature image and each HOG feature image in the neighborhood thereof, obtaining the direction deviation value of the overall gradient direction corresponding to each HOG feature image in the HOG feature image and each HOG feature image in the neighborhood thereof, reflecting the intra-neighborhood identification of the HOG feature image according to the direction deviation value, and adjusting the importance degree in the local range of the subsequent HOG feature image according to the direction deviation value, wherein the obtaining of the second importance degree of each HOG feature image according to the direction deviation value and the gradient amplitude value comprises the following steps: acquiring the sum of absolute values of gradient amplitude difference values of the HOG characteristic image and the HOG characteristic image in the adjacent region in all corresponding gradient directions; obtaining the sum of absolute values of gradient amplitude values and the product of the absolute value of the direction deviation value of the whole gradient direction of the HOG characteristic image and the corresponding HOG characteristic image in the adjacent area; and taking the sum of products of the sum of absolute values of gradient amplitude differences corresponding to all HOG feature images in the neighborhood of the HOG feature images and the direction deviation value of the whole gradient direction as a second important degree of the HOG feature images, wherein a calculation formula of the second important degree is as follows:
In the method, in the process of the invention,
representing HOG feature images
Is of a second importance;
It should be noted that, in this embodiment, the neighborhood size is taken as 8 neighborhood, so that the 8 neighborhood range of one HOG feature image has similarity in the face image, the gradient direction and the corresponding gradient amplitude of the adjacent HOG feature image are similar, and when the difference value between the HOG feature image and the rest HOG feature images in the neighborhood range is large in the gradient direction and the corresponding gradient amplitude, it is indicated that the HOG feature image has high identification, that is, the HOG feature image is reflected by the formula
The significant extent of the gradient in its neighborhood is the second important extent.
S4, acquiring the comprehensive importance degree of each HOG characteristic image;
specifically, acquiring the comprehensive importance degree of each HOG feature image according to the first importance degree and the second importance degree includes: and setting weights of the first importance degree and the second importance degree to be 0.5, and adding the first importance degree corresponding to the HOG feature image multiplied by 0.5 and the second importance degree corresponding to the HOG feature image multiplied by 0.5 to obtain the comprehensive importance degree of the HOG feature image.
In addition, the problem that light is affected in a scene is acquired when face information is acquired, and three-dimensional features of five sense organs are acquired, wherein shadow conditions and light reflection conditions exist on faces of people more or less, the factors can cause incomplete features of partial areas in the acquired HOG feature images, such as shadows on one side of a nose area, further, the features of all gradient amplitude distribution in the HOG feature images are weakened, further, the second importance degree of the HOG feature images due to the partial features is reduced, and finally, when the change degree is acquired later, the transformation degree of the important HOG feature images is smaller, so that corresponding adjustment of the acquired HOG feature images is needed.
Therefore, the face data encryption transmission method for terminal identity authentication of the invention further comprises the following steps: acquiring each HOG characteristic image and a target HOG characteristic image symmetrical about the vertical center line of the face image; according to the overall gradient direction of each HOG feature image and the corresponding target HOG feature image, the direction symmetry degree of each HOG feature image in the neighborhood of the HOG feature image and the HOG feature image at the symmetrical position in the neighborhood of the corresponding target HOG feature image is obtained, as shown in figure 2 
And HOG feature images
Symmetrical; according to the direction deviation value of the overall gradient direction of each HOG characteristic image in the neighborhood of the HOG characteristic image and the HOG characteristic image in the symmetrical position in the neighborhood of each HOG characteristic image and the corresponding target HOG characteristic image, acquiring the overall gradient direction symmetry degree of the HOG characteristic image and the corresponding target HOG characteristic image in the neighborhood range; according to HOG specialJudging whether to adjust the neighborhood size according to the sum of the direction deviation values of the overall gradient directions of the feature images and all HOG feature images in the neighborhood of the feature images, and obtaining the final neighborhood size corresponding to each HOG feature image; normalizing according to the length of the final neighborhood size corresponding to each HOG characteristic image to obtain the necessity degree of the HOG characteristic image; and adjusting the comprehensive importance degree of each corresponding HOG characteristic image according to the corresponding necessity degree of the HOG characteristic image, the overall gradient direction symmetry degree and the sum of the gradient magnitudes of the HOG characteristic image and the corresponding target HOG characteristic image and the adjustment value to obtain the final comprehensive importance degree.
Wherein obtaining the final neighborhood size comprises: when the sum of the direction deviation values of the whole gradient directions of the HOG characteristic images and all HOG characteristic images in the neighborhood of the HOG characteristic images is smaller than a preset sum threshold, increasing the neighborhood size of the HOG characteristic images to obtain an increased target neighborhood; acquiring the sum of the direction deviation values of the overall gradient directions of the HOG feature images and each HOG feature image in the target neighborhood, and taking the size of the corresponding target domain when the sum of the direction deviation values of the overall gradient directions of the HOG feature images and each HOG feature image in the target neighborhood is larger than or equal to a preset sum threshold as the final neighborhood size, wherein the HOG feature image in the embodiment 
The direction deviation values of the overall gradient directions of the characteristic images in the 3*3 neighborhood range are summed, namely
，
A sum of direction deviation values representing the overall gradient direction of the HOG feature image and all feature images within its 3*3 neighborhood,
representing HOG feature images
HOG feature images in its neighborhood
The present embodiment sets the sum threshold as the direction deviation value of the overall gradient direction of
(threshold value)
The practitioner can adjust parameters such as the precision setting size and the like when the user converts the resolution of the terminal to HOG face data model according to the resolution of the terminal used for acquiring the face image, and the experience value is 500) if
Step 2 is increased by the length of the neighborhood range to obtain the increased neighborhood range, namely the increased neighborhood range is
Is used for calculating HOG characteristic images
Adding up the direction deviation values of the overall gradient directions of the characteristic images in 5*5 neighborhood range
If (if)
Then continue to increase the neighborhood range until
In this case, the neighborhood range is
Will then
As the final neighborhood size.
The obtaining the necessity degree of the HOG characteristic image comprises the following steps: acquiring the length of the corresponding final neighborhood size and the length of the minimum neighborhood size in the lengths of the final neighborhood sizes corresponding to all HOG characteristic images; acquiring a target neighborhood size corresponding to the normalized HOG feature image according to the length of the corresponding final neighborhood size, the length of the minimum neighborhood size and the length of the final neighborhood size corresponding to each HOG feature image; regarding the length corresponding to the target neighborhood size as the necessity of the HOG feature image, it should be noted that, in the case of considering face recognition, the recognition is based on local features, and if the corresponding adjustment is performed only by the amplitude difference of the symmetrical positions, then the adjustment is performed on a part of unimportant regions, such as the hair, the beard, the cheek, etc., so that the difference between the unimportant region and the important region is weakened, therefore, the necessity of the adjustment of the HOG feature image needs to be performed according to the identification property represented by the gradient direction features of the local region, and specifically, the necessity of the HOG feature image has a calculation formula:
In the method, in the process of the invention,
representing HOG feature images
Is the degree of necessity of (1);
it should be noted that, when the accumulated value of the overall gradient direction deviation values in the neighborhood range where the HOG feature image is located reaches the threshold value, the larger the required range is, the worse the identification of the local position where the HOG feature image is located is, the lower the importance degree of the local position in face recognition is, that is, the smaller the necessary degree of adjustment is needed, and then, the normalization formula is the formula in the prior art, which is not described in detail in this embodiment.
The calculation formula of the direction symmetry degree of each HOG characteristic image in the neighborhood of the HOG characteristic image and the HOG characteristic image at the symmetrical position in the neighborhood of the corresponding target HOG characteristic image is as follows:
in the method, in the process of the invention,
representing HOG feature images
HOG feature images in the neighborhood of (2)
And corresponding target HOG characteristic image
In the neighborhood of (a) with HOG feature images
The degree of directional symmetry of the position-symmetric HOG feature image;
in this embodiment, the vertical center line of the face image is taken as a symmetry line, that is, the face image is bilaterally symmetric about the vertical center line, the embodiment specifies that the horizontal direction is 0 ° to the right, and therefore, the vertical center line is 90 °, and the bilateral symmetry is taken as an example of 135 ° and 45 ° and is symmetrical about the center line, and therefore, the closer to 90 ° the half of the sum of the magnitudes of the two gradient directions, the higher the degree of bilateral symmetry of the two gradient directions is, and the addition of 1 in the formula is to prevent the denominator from being 0.
The step of obtaining the overall gradient direction symmetry degree of the HOG characteristic image and the corresponding target HOG characteristic image in the neighborhood range comprises the following steps: acquiring a direction deviation value of the overall gradient direction of each HOG characteristic image in the neighborhood of the HOG characteristic image and the sum value of the ratio of the direction symmetry degree of each HOG characteristic image in the neighborhood of the HOG characteristic image and the HOG characteristic image in the symmetrical position in the neighborhood of the corresponding target HOG characteristic image; and taking the sum value as the overall gradient direction symmetry degree of the HOG characteristic image and the corresponding target HOG characteristic image in the neighborhood range, wherein the calculation formula of the overall gradient direction symmetry degree is as follows:
In the method, in the process of the invention,
representing the overall gradient direction symmetry degree of the HOG characteristic image and the corresponding target HOG characteristic image in the neighborhood range;
the degree of symmetry of the direction is to be noted
The bigger the HOG feature image is
And corresponding target HOG characteristic image
The more bilateral symmetry the HOG features in the local range, i.e. the greater the degree of directional symmetry, the more the overall gradient directional deviation value
The bigger the two points reflected on the bilateral symmetry position in the face image do not necessarily completely correspond to the actual symmetry position, but the symmetry position still keeps the characteristic that the whole gradient direction is bilateral symmetry, and because the local similarity, the face area corresponding to the 8 neighborhood range is smaller in practice, therefore, the whole gradient direction should be very similar, so the deviation value of the whole gradient direction is larger, the confidence of the symmetry judgment of the local position is very low, and the direction deviation value is very low 
Degree of symmetry with overall gradient direction
Is a negative correlation.
Wherein, the adjustment value calculation formula of HOG feature image:
in the method, in the process of the invention,
representing HOG feature images
Is a set of the adjustment values of (2);
when the sum of the gradient magnitudes of the HOG feature images is greater than or equal to the sum of the gradient magnitudes of the corresponding target HOG feature images, the adjustment value is 1, that is, adjustment is not needed, when the sum of the gradient magnitudes of the HOG feature images is smaller than the sum of the gradient magnitudes of the corresponding target HOG feature images, adjustment is needed, so that when the symmetry degree of the whole gradient directions of the HOG feature images and the corresponding target HOG feature images in the neighborhood range is greater, the symmetry of the HOG feature images and the corresponding target HOG feature images in the symmetry position is greater, the symmetry is greater, the corresponding adjustment value is greater, and when the necessity is greater, the ratio of the sum of the gradient magnitudes of the HOG feature images to the sum of the gradient magnitudes of the corresponding target HOG feature images is smaller, the corresponding adjustment value is greater when the ratio is smaller.
The method comprises the steps of obtaining comprehensive importance degrees of each HOG characteristic image according to an adjustment value, a first importance degree and a second importance degree, wherein a calculation formula of the comprehensive importance degrees is as follows:
in the method, in the process of the invention,
representing HOG feature images
Is a comprehensive importance level of (2);
the first importance level and the second importance level are positively correlated with the importance level, and in the present embodiment, the importance of the first importance level and the second importance level is considered to be equal, and thus the same is set for the first importance level and the second importance levelThe weight value is 0.5, and secondly,
the larger the image, the more obvious the self-identification of the reflected HOG feature image,
the bigger the HOG feature image is, the bigger the local identification of the HOG feature image in the neighborhood range is, the bigger the local identification is, the stronger the identification of the HOG feature image is, namely the higher the comprehensive importance degree is, the more encryption is needed, the light influence when the terminal collects the face is considered, so that the important area features are smaller in amplitude when the HOG feature image is used for recognizing the face data, therefore, the adjustment value of the HOG feature image with smaller amplitude is obtained by calculating the difference between the HOG feature images in the symmetrical positions, and the too small comprehensive importance degree of the HOG feature image in the important positions is avoided.
S5, acquiring a final HOG characteristic image of each image block;
specifically, acquiring an initial transformation gradient amplitude value of each gradient direction according to a gradient amplitude value average value of each gradient direction and a gradient amplitude value of each gradient direction in each HOG characteristic image; obtaining a target gradient amplitude value of each gradient direction in each HOG characteristic image by utilizing the gradient amplitude value of each gradient direction and the initial transformation gradient amplitude value of each gradient direction in each HOG characteristic image, obtaining a transformation coefficient of each gradient direction according to the initial transformation gradient amplitude value and the target gradient amplitude value of each gradient direction, obtaining a final transformation gradient amplitude value of each gradient direction of the HOG characteristic image according to the transformation coefficient and the initial transformation gradient amplitude value of each gradient direction and the comprehensive importance degree of the HOG characteristic image, and obtaining a final HOG characteristic image of each image block according to all final transformation gradient amplitude values.
Because the gradient direction histogram has a large degree of difference between partial magnitudes, the partial gradient magnitudes with large magnitudes need to be transformed, so that the part with large gradient magnitude difference in each HOG characteristic image is eliminated, the direction gradient histograms of all HOG characteristic images are similar, the gradient magnitudes in each gradient direction in the HOG characteristic image and the gradient magnitude mean value in the corresponding gradient direction in the HOG characteristic image are differed to obtain the initial transformed gradient magnitude in each gradient direction, then the gradient magnitude in each gradient direction is subtracted by the initial transformed gradient magnitude in each gradient direction to obtain the gradient after transformation in each direction, namely the target gradient magnitude, the duty ratio of the initial transformed gradient magnitude in the transformed target gradient magnitude is calculated and obtained and is used as a transformation coefficient, the transformation coefficient is positive and negative because the gradient is vector, and finally the final transformed gradient is calculated, wherein the calculation formula of the final transformed gradient is as follows:
In the method, in the process of the invention,
first representing HOG feature image
Final transformed gradient magnitudes for each gradient direction;
it should be noted that, because the prior art performs preliminary normalization on the directional gradient histograms, but at this time, the transformation rule between the directional gradient histograms corresponding to each HOG feature image is identical, the obtained key is too simple, and although the attack mode of statistical analysis of the histogram can be resisted to a certain extent, the resistance capability to the attack mode of brute force cracking is weaker, and meanwhile, the security is also lower, so in this embodiment, the comprehensive importance degree is used as the adjustment value of the transformation coefficient, the transformation coefficient is adjusted, so that the transformation gradient amplitude is also adjusted, and the security of the final HOG feature image is higher.
Based on a calculation formula of the final change gradient of each gradient direction, all final change gradient amplitudes of each HOG characteristic image can be obtained, so that the change of the direction gradients of different degrees of each HOG characteristic image with different comprehensive importance degrees is realized, the amplitude difference among a plurality of direction gradients in the same HOG characteristic image is reduced, and meanwhile, the identification of the HOG characteristic image in the neighborhood range is also reduced.
S6, acquiring an encrypted image and transmitting the encrypted image;
specifically, an encrypted image is obtained according to all the final HOG feature images, and the encrypted image, the transformation coefficient corresponding to each final HOG feature image and the comprehensive importance degree are transmitted.
And the conversion coefficient and the comprehensive importance degree corresponding to each final HOG characteristic image are stored as the key of each corresponding final HOG characteristic image, so that the encrypted transmission of the face image is realized.
Further comprises: decrypting the encrypted data, the decrypting step comprising: acquiring the gradient before transformation of each gradient direction of the HOG characteristic image according to each final HOG characteristic image, the corresponding transformation coefficient and the comprehensive importance degree; acquiring HOG characteristic images according to gradients before transformation of each gradient direction of the HOG characteristic images; obtaining a face image according to all HOG characteristic images, and obtaining gradient before transformation of each gradient direction
Wherein, the method comprises the steps of, wherein,
represent the first
Transform coefficients for each gradient direction;
represents the comprehensive importance of the HOG feature image,
first representing final HOG feature image
The final transformation gradient amplitude of each gradient direction is obtained based on the own HOG feature and the local HOG feature in the adjacent domain, namely, the comprehensive importance degree obtained based on the second importance degree and the first importance degree, so that when one HOG feature image is cracked, the cracking result of the next HOG feature image is wrong, and therefore, the avalanche effect exists during cracking, and the safety of encrypted data is higher.
The invention relates to a face data encryption transmission method for terminal identity authentication, which is characterized in that the gradient directions of areas corresponding to different facial organs in a face image are different, so the face image is divided into a plurality of image blocks, then HOG characteristic images representing each image block are obtained, the first importance degree of the HOG characteristic images is reflected by the standard deviation of the gradient amplitude values of the HOG characteristic images corresponding to all gradient directions, the larger the standard deviation is, the larger the gradient amplitude difference is, the corresponding first importance degree is also, namely the importance degree of the HOG characteristic images is reflected by the first importance degree, then the second importance degree of the HOG characteristic images is reflected by the direction deviation values and the gradient amplitude values of the HOG characteristic images and all HOG characteristic images in the neighborhood, namely, the second important degree reflects the importance degree of the HOG characteristic image and the HOG characteristic of the HOG characteristic image in the field, so that the HOG characteristic image is combined with the first importance degree of the HOG characteristic image, the HOG characteristic image and the HOG characteristic of the HOG characteristic image in the field are combined with the second importance degree of the HOG characteristic image and the HOG characteristic of the HOG characteristic image in the field and the comprehensive importance degree of the HOG characteristic image are comprehensively evaluated, so that when the HOG characteristic image is attacked, the HOG characteristic image is cracked, the decoding results of the subsequent HOG characteristic images are wrong, so that the decoding has avalanche effect, so that the safety of the encrypted data obtained by encryption based on the comprehensive importance degree is higher, and the comprehensive importance degree is characterized by the significance of the gradient direction in the HOG characteristic images, and the direction gradient of the region corresponding to the face organ is different from the forehead and cheek region, so that the HOG characteristic images corresponding to the face organ region have significant gradient directions, if the magnitude of the significant gradient direction is transformed, so that the gradient magnitude of each HOG characteristic image in the whole image is similar, the difference between all HOG characteristic images of the whole image is small, even if stored encrypted data is stolen, the stored encrypted data cannot be cracked, the gradient of the HOG characteristic image is transformed according to the comprehensive importance degree and the transformation coefficient of each gradient direction to obtain the final transformed gradient magnitude, and the final HOG characteristic image of each image block is obtained according to the final transformed gradient magnitude, so that the gradient magnitudes corresponding to all the final HOG characteristic images are similar, namely, the comprehensive importance degree of the HOG characteristic images is utilized to be different, encryption of different HOG characteristic images is realized, and the security of transmitted data is further improved.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.
Claims (10)
1. The face data encryption transmission method for terminal identity authentication is characterized by comprising the following steps:
acquiring a face image, and dividing the face image to obtain a plurality of identical image blocks;
acquiring HOG characteristic images of each image block, and taking standard deviations of gradient amplitudes of all gradient directions in each HOG characteristic image as a first importance degree of each HOG characteristic image;
acquiring the overall gradient direction of each HOG characteristic image according to the horizontal gradient amplitude component and the vertical gradient amplitude component of the gradient amplitudes of all gradient directions in each HOG characteristic image; acquiring a second important degree of each HOG characteristic image according to the direction deviation value of the overall gradient direction of each HOG characteristic image and the overall gradient direction of each HOG characteristic image in the neighborhood and the gradient amplitude value difference value of each HOG characteristic image and each HOG characteristic image in the neighborhood in the corresponding gradient direction;
acquiring the comprehensive importance degree of each HOG characteristic image according to the first importance degree and the second importance degree;
Acquiring initial transformation gradient amplitude values of each gradient direction according to the gradient amplitude value average value of each gradient direction and the gradient amplitude value of each gradient direction in each HOG characteristic image; obtaining a target gradient amplitude value of each gradient direction in each HOG characteristic image by utilizing the gradient amplitude value of each gradient direction in each HOG characteristic image and the initial transformation gradient amplitude value of each gradient direction;
obtaining a transformation coefficient of each gradient direction according to the initial transformation gradient amplitude of each gradient direction and the target gradient amplitude, obtaining a final transformation gradient amplitude of each gradient direction of the HOG characteristic image according to the transformation coefficient of each gradient direction, the initial transformation gradient amplitude and the comprehensive importance degree of the HOG characteristic image, and obtaining a final HOG characteristic image of each image block according to all final transformation gradient amplitudes;
and obtaining an encrypted image according to all the final HOG characteristic images, and transmitting the encrypted image, the transformation coefficient corresponding to each final HOG characteristic image and the comprehensive importance degree.
2. The method for encrypting and transmitting face data for terminal identity authentication according to claim 1, further comprising:
Acquiring each HOG characteristic image and a target HOG characteristic image symmetrical about the vertical center line of the face image;
according to the overall gradient direction of each HOG characteristic image and the corresponding target HOG characteristic image, obtaining the direction symmetry degree of each HOG characteristic image in the neighborhood of the HOG characteristic image and the HOG characteristic image at the symmetrical position in the neighborhood of the corresponding target HOG characteristic image;
according to the direction deviation value of the overall gradient direction of each HOG characteristic image in the neighborhood of the HOG characteristic image and the HOG characteristic image in the symmetrical position in the neighborhood of each HOG characteristic image and the corresponding target HOG characteristic image, acquiring the overall gradient direction symmetry degree of the HOG characteristic image and the corresponding target HOG characteristic image in the neighborhood range;
judging whether to carry out neighborhood size adjustment according to the sum value of the direction deviation values of the whole gradient directions of the HOG characteristic images and all HOG characteristic images in the neighborhood of the HOG characteristic images, and obtaining the final neighborhood size corresponding to each HOG characteristic image;
normalizing according to the length of the final neighborhood size corresponding to each HOG characteristic image to obtain the necessity degree of the HOG characteristic image;
Acquiring an adjustment value of each HOG characteristic image according to the corresponding necessity degree of the HOG characteristic image, the symmetry degree of the whole gradient direction and the sum of the gradient magnitudes of the HOG characteristic image and the corresponding target HOG characteristic image;
and adjusting the comprehensive importance degree of each corresponding HOG characteristic image according to the adjustment value to obtain the final comprehensive importance degree.
3. The method for encrypting and transmitting face data for terminal identity authentication according to claim 2, wherein obtaining the final neighborhood size comprises:
when the sum of the direction deviation values of the whole gradient directions of the HOG characteristic images and all HOG characteristic images in the neighborhood of the HOG characteristic images is smaller than a preset sum threshold, gradually increasing the neighborhood size of the HOG characteristic images to obtain an increased target neighborhood;
and acquiring the sum value of the direction deviation values of the overall gradient directions of the HOG characteristic images and each HOG characteristic image in the target neighborhood, and taking the size of the corresponding target neighborhood as the final neighborhood size when the sum value of the direction deviation values of the overall gradient directions of the HOG characteristic images and each HOG characteristic image in the target neighborhood is larger than or equal to a preset sum value threshold.
4. The face data encryption transmission method for terminal identity authentication according to claim 2, wherein obtaining the necessity of the HOG feature image comprises:
Acquiring the length of the corresponding final neighborhood size and the length of the minimum neighborhood size in the lengths of the final neighborhood sizes corresponding to all HOG characteristic images;
obtaining a target neighborhood size corresponding to the normalized HOG feature image according to the length of the corresponding final neighborhood size, the length of the minimum neighborhood size and the length of the final neighborhood size corresponding to each HOG feature image;
and taking the length corresponding to the target neighborhood size as the necessity of the HOG characteristic image.
5. The face data encryption transmission method for terminal identity authentication according to claim 2, wherein the calculation formula of the directional symmetry degree of each HOG feature image in the neighborhood of the HOG feature image and the HOG feature image at the symmetrical position in the neighborhood of the corresponding target HOG feature image:
in the method, in the process of the invention,
representing HOG feature image +.>
HOG feature images in the neighborhood of +.>
And corresponding target HOG feature image +.>
In the neighborhood of (2) with the HOG feature image +.>
The degree of directional symmetry of the position-symmetric HOG feature image;
6. The method for encrypting and transmitting face data for terminal identity authentication according to claim 2, wherein obtaining the degree of symmetry of the whole gradient directions of the HOG feature image and the corresponding target HOG feature image in the neighborhood range comprises:
acquiring a direction deviation value of the overall gradient direction of each HOG characteristic image in the neighborhood of the HOG characteristic image and the sum value of the ratio of the direction symmetry degree of each HOG characteristic image in the neighborhood of the HOG characteristic image and the HOG characteristic image in the symmetrical position in the neighborhood of the corresponding target HOG characteristic image;
and taking the sum value as the overall gradient direction symmetry degree of the HOG characteristic image and the corresponding target HOG characteristic image in the neighborhood range.
7. The face data encryption transmission method for terminal identity authentication according to claim 2, wherein the calculation formula of the adjustment value of each HOG feature image is:
in the method, in the process of the invention,
representing HOG feature image +.>
Is a set of the adjustment values of (2);
8. The face data encryption transmission method for terminal identity authentication according to claim 1, wherein obtaining the second importance degree of each HOG feature image comprises:
acquiring the sum of absolute values of gradient amplitude difference values of the HOG characteristic image and the HOG characteristic image in the adjacent region in all corresponding gradient directions;
and taking the sum of products of the sum of absolute values of gradient amplitude differences corresponding to all HOG feature images in the neighborhood of the HOG feature images and the direction deviation value of the whole gradient direction as a second important degree of the HOG feature images.
9. The method for encrypting and transmitting face data for terminal identity authentication according to claim 1, wherein obtaining final transformed gradient magnitudes for each gradient direction of the HOG feature image comprises:
obtaining the product of the transformation coefficient and the comprehensive importance degree;
and obtaining the ratio of the initial transformation gradient amplitude corresponding to each gradient direction in the HOG characteristic image to the product of the transformation coefficient and the comprehensive importance degree, and taking the ratio as the final transformation gradient amplitude of each gradient direction of the HOG characteristic image.
10. The face data encryption transmission method for terminal identity authentication according to claim 1, further comprising: decrypting the encrypted data, the decrypting step comprising:
acquiring the gradient before transformation of each gradient direction of the HOG characteristic image according to each final HOG characteristic image, the corresponding transformation coefficient and the comprehensive importance degree;
acquiring HOG characteristic images according to gradients before transformation of each gradient direction of the HOG characteristic images;
and obtaining a face image according to all the HOG characteristic images.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310043171.1A CN115776410B (en) | 2023-01-29 | 2023-01-29 | Face data encryption transmission method for terminal identity authentication |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310043171.1A CN115776410B (en) | 2023-01-29 | 2023-01-29 | Face data encryption transmission method for terminal identity authentication |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115776410A CN115776410A (en) | 2023-03-10 |
| CN115776410B true CN115776410B (en) | 2023-05-02 |
Family
ID=85393743
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310043171.1A Active CN115776410B (en) | 2023-01-29 | 2023-01-29 | Face data encryption transmission method for terminal identity authentication |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115776410B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116541872B (en) * | 2023-07-07 | 2024-04-09 | 深圳奥联信息安全技术有限公司 | Data information safety transmission method and system |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105550646A (en) * | 2015-12-08 | 2016-05-04 | 中山大学 | Generalized illumination invariant face feature description method based on logarithmic gradient histogram |
| CN109712143A (en) * | 2018-12-27 | 2019-05-03 | 北京邮电大学世纪学院 | A kind of Fast image segmentation method based on super-pixel multiple features fusion |
| CN111008589A (en) * | 2019-12-02 | 2020-04-14 | 杭州网易云音乐科技有限公司 | Face key point detection method, medium, device and computing equipment |
| CN111832405A (en) * | 2020-06-05 | 2020-10-27 | 天津大学 | Face recognition method based on HOG and depth residual error network |
| WO2021139171A1 (en) * | 2020-07-28 | 2021-07-15 | 平安科技（深圳）有限公司 | Facial enhancement based recognition method, apparatus and device, and storage medium |
| WO2021227349A1 (en) * | 2020-05-11 | 2021-11-18 | 华南理工大学 | Front-end facial image encryption and recognition method for biometric privacy protection |
| CN113762970A (en) * | 2021-05-17 | 2021-12-07 | 腾讯科技（深圳）有限公司 | Data processing method and device, computer readable storage medium and computer equipment |
| CN113766085A (en) * | 2021-05-17 | 2021-12-07 | 腾讯科技（深圳）有限公司 | Image processing method and related device |
| CN114419716A (en) * | 2022-01-26 | 2022-04-29 | 北方工业大学 | Calibration method for face key point calibration of face image |
| CN115620117A (en) * | 2022-12-20 | 2023-01-17 | 吉林省信息技术研究所 | Face information encryption method and system for network access authority authentication |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2624329A (en) * | 2021-07-01 | 2024-05-15 | Deka Products Lp | Surveillance data filtration techniques |
-
2023
- 2023-01-29 CN CN202310043171.1A patent/CN115776410B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105550646A (en) * | 2015-12-08 | 2016-05-04 | 中山大学 | Generalized illumination invariant face feature description method based on logarithmic gradient histogram |
| CN109712143A (en) * | 2018-12-27 | 2019-05-03 | 北京邮电大学世纪学院 | A kind of Fast image segmentation method based on super-pixel multiple features fusion |
| CN111008589A (en) * | 2019-12-02 | 2020-04-14 | 杭州网易云音乐科技有限公司 | Face key point detection method, medium, device and computing equipment |
| WO2021227349A1 (en) * | 2020-05-11 | 2021-11-18 | 华南理工大学 | Front-end facial image encryption and recognition method for biometric privacy protection |
| CN111832405A (en) * | 2020-06-05 | 2020-10-27 | 天津大学 | Face recognition method based on HOG and depth residual error network |
| WO2021139171A1 (en) * | 2020-07-28 | 2021-07-15 | 平安科技（深圳）有限公司 | Facial enhancement based recognition method, apparatus and device, and storage medium |
| CN113762970A (en) * | 2021-05-17 | 2021-12-07 | 腾讯科技（深圳）有限公司 | Data processing method and device, computer readable storage medium and computer equipment |
| CN113766085A (en) * | 2021-05-17 | 2021-12-07 | 腾讯科技（深圳）有限公司 | Image processing method and related device |
| CN114419716A (en) * | 2022-01-26 | 2022-04-29 | 北方工业大学 | Calibration method for face key point calibration of face image |
| CN115620117A (en) * | 2022-12-20 | 2023-01-17 | 吉林省信息技术研究所 | Face information encryption method and system for network access authority authentication |
Non-Patent Citations (1)
| Title |
|---|
| 融合人脸特征与密码算法的身份认证系统;赖韬等;《电讯技术》(第第202209期期);I284-I291 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115776410A (en) | 2023-03-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Mu et al. | Shape and structural feature based ear recognition | |
| CN108921191B (en) | Multi-biological-feature fusion recognition method based on image quality evaluation | |
| CN115776410B (en) | Face data encryption transmission method for terminal identity authentication | |
| CN111523344B (en) | Human body living body detection system and method | |
| CN113055377A (en) | Network security protection system based on authority authentication | |
| CN111507206A (en) | Finger vein identification method based on multi-scale local feature fusion | |
| Sushama et al. | Face recognition using DRLBP and SIFT feature extraction | |
| Hofbauer et al. | Image metric-based biometric comparators: A supplement to feature vector-based hamming distance? | |
| CN110163123B (en) | Fingerprint finger vein fusion identification method based on single near-infrared finger image | |
| CN106599841A (en) | Full face matching-based identity verifying method and device | |
| CN106650657A (en) | Authentication method and device based on full face binary matching | |
| CN110503697B (en) | Iris feature hiding method based on random noise mechanism | |
| US7436999B2 (en) | Data analysis device and data recognition device | |
| CN115861034B (en) | Wireless routing data intelligent management system | |
| CN116778562A (en) | Face verification method, device, electronic equipment and readable storage medium | |
| CN102087705B (en) | Iris identification method based on blanket dimension and lacunarity | |
| CN112651319B (en) | Video detection method and device, electronic equipment and storage medium | |
| Adamović et al. | Information analysis of iris biometrics for the needs of cryptology key extraction | |
| CN112070956A (en) | Identity recognition system based on face image | |
| CN110502996B (en) | Dynamic identification method for fuzzy finger vein image | |
| CN110956098B (en) | Image processing method and related equipment | |
| CN114120398A (en) | Non-inductive intelligent sign-in system and method for outdoor conference | |
| Kim et al. | Implementation and enhancement of GMM face recognition systems using flatness measure | |
| Paunwala et al. | Dct watermarking approach for security enhancement of multimodal system | |
| Patil et al. | A comparative study of feature extraction approaches for an efficient iris recognition system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |