CN111259806B - Face area identification method, device and storage medium - Google Patents
Face area identification method, device and storage medium Download PDFInfo
- Publication number
- CN111259806B CN111259806B CN202010050220.0A CN202010050220A CN111259806B CN 111259806 B CN111259806 B CN 111259806B CN 202010050220 A CN202010050220 A CN 202010050220A CN 111259806 B CN111259806 B CN 111259806B
- Authority
- CN
- China
- Prior art keywords
- feature
- objects
- feature object
- affinity
- candidate
- 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
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000005457 optimization Methods 0.000 claims abstract description 42
- 230000006870 function Effects 0.000 claims description 29
- 238000004590 computer program Methods 0.000 claims description 24
- 230000010076 replication Effects 0.000 claims description 10
- 230000035772 mutation Effects 0.000 claims description 9
- 238000004422 calculation algorithm Methods 0.000 abstract description 10
- 230000006872 improvement Effects 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 230000001965 increasing effect Effects 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 2
- 230000001815 facial effect Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000003491 array Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000010429 evolutionary process Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/161—Detection; Localisation; Normalisation
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a face area identification method, a device and a storage medium, wherein the method comprises the following steps: acquiring an image to be identified, and randomly selecting a plurality of elliptic areas as initial characteristic objects; calculating the affinity of the initial feature object; selecting a plurality of initial feature objects according to the affinity to perform copying operation and feature optimization operation to obtain a plurality of candidate feature objects; performing iterative selection on candidate feature objects according to a preset selection rule, so as to select a part of candidate feature objects to perform the next iteration of the copying operation and the feature optimization operation, and simultaneously determining the current optimal feature objects; and stopping iteration selection when the iteration times reach a preset first threshold value or when the difference value of affinities of two previous and subsequent iterations corresponding to the current optimal feature object is smaller than a preset second threshold value, and taking the current optimal feature object as a face area. The invention can accurately and rapidly locate the face region through the evolutionary selection algorithm, and is simple and easy to operate.
Description
Technical Field
The present invention relates to the field of face recognition technologies, and in particular, to a face region recognition method, device, and storage medium.
Background
The face recognition system is an emerging biological recognition technology, is a high-precision technology which is an attack of the current international technology field, and has wide development prospect. Face area recognition extraction has been a difficult problem in face recognition systems. The existing face region identification methods generally have the problems of complex identification algorithm, low identification speed and low accuracy.
Disclosure of Invention
The embodiment of the invention aims to provide a face region identification method, a face region identification device and a storage medium, which can accurately and rapidly locate a face region through an evolutionary selection algorithm, and are simple and easy to operate.
In order to achieve the above object, an embodiment of the present invention provides a face region recognition method, including the following steps:
acquiring an image to be identified, and randomly selecting a plurality of elliptic areas as initial characteristic objects;
calculating the affinity of the initial feature object;
selecting a plurality of initial feature objects to copy according to the affinity degree to obtain a plurality of feature objects to be optimized;
performing feature optimization operation on the feature object to be optimized to obtain a plurality of candidate feature objects;
performing iterative selection on the candidate feature objects according to a preset selection rule, so as to select a part of the candidate feature objects to perform the next iteration of the copying operation and the feature optimization operation, and simultaneously determining the current optimal feature objects;
And stopping iteration selection when the iteration times reach a preset first threshold or when the difference value of affinities of two previous and subsequent iterations corresponding to the current optimal feature object is smaller than a preset second threshold, and taking the current optimal feature object as a face area.
Preferably, the acquiring the image to be identified randomly selects a plurality of elliptical areas as the initial feature objects, and specifically includes:
acquiring an image to be identified, and establishing a coordinate axis by taking the lower left corner of the image to be identified as a circle center; the width of the image to be identified is x, the length of the image to be identified is y, and the sitting mark of the image to be identified is (x, y);
setting the coding of the elliptical area as (i, j, w, h); wherein (i, j) is the lower left angular coordinate of the oval-shaped region circumscribed rectangle, w is the minor axis of the oval-shaped region, and h is the major axis of the oval-shaped region;
randomly selecting a plurality of elliptical areas as initial feature objects by adopting an initialization mode of comprehensive random and priori knowledge according to constraint conditions; wherein, the constraint condition is that i is more than or equal to 0 and less than or equal to x, j is more than or equal to 0 and less than or equal to y, i+w is more than or equal to 0 and less than or equal to x, and j+h is more than or equal to 0 and less than or equal to y.
Preferably, the calculating the affinity of the initial feature object specifically includes:
calculating the density of the gray scale fluctuation times of the initial characteristic object;
Calculating the gray level difference of the peak mean value and the trough mean value of the initial feature object;
and obtaining the affinity of the initial characteristic object according to the density and the gray level difference.
Preferably, the selecting a plurality of initial feature objects according to the affinity to perform the copying operation to obtain a plurality of feature objects to be optimized specifically includes:
the initial feature objects are arranged in a descending order according to the affinity degree, and a plurality of initial feature objects arranged in front are selected;
copying the selected plurality of initial feature objects to obtain a plurality of feature objects to be optimized; wherein the copy operation isC j For the initial feature object, I j Q is an element value of 1 j A dimension vector, j is more than or equal to 1 and less than or equal to k,q j for the initial feature object C j The scale after replication; int []K is the number of the initial feature objects selected as the upper rounding function, N c For the number of the feature objects to be optimized obtained after the copying operation, k is more than or equal to 1 and less than N c ；f(C j ) As a function of affinity.
Preferably, the feature optimization operation is performed on the feature object to be optimized to obtain a plurality of candidate feature objects, which specifically includes:
selecting one or more feature objects to be optimized according to a preset first probability, and acquiring codes of the feature objects to be optimized;
Performing mutation operation on one or more bits of the codes of the feature object to be optimized; wherein the code comprises four bits, i, j, w, h respectively;
and integrating the mutated feature object to be optimized with the non-mutated feature object to be optimized to obtain a plurality of candidate feature objects.
Preferably, the iterative selection of the candidate feature objects according to a preset selection rule is performed, so that in the iteration of selecting a part of candidate feature objects for performing the next copy operation and feature optimization operation, the current optimal feature object is determined, and the method specifically includes:
sorting the candidate feature objects in a descending order according to the affinity, and selecting a plurality of candidate feature objects arranged in front;
sequentially comparing the affinity of two adjacent candidate feature objects;
if the affinity of the candidate feature object at the current high level is smaller than that of the candidate feature object at the current low level, enabling the candidate feature object at the current low level to be a value object of the candidate feature object at the current high level in the iteration of the next copying operation and feature optimization operation;
if the affinity of the candidate feature object at the current high level is greater than or equal to that of the candidate feature object at the current low level and the affinity of the candidate feature object at the current high level is not the highest, making the candidate feature object at the current low level be a value object of the candidate feature object at the current high level in the iteration of the next copying operation and feature optimizing operation;
If the affinity of the candidate feature object at the current high level is greater than or equal to that of the candidate feature object at the current low level and the affinity of the candidate feature object at the current high level is the highest, making the candidate feature object at the current high level be a value object of the candidate feature object at the current high level in the iteration of the next copying operation and feature optimizing operation;
and after the next iteration value object of each candidate feature object is determined, taking the candidate feature object with the highest affinity as the current optimal feature object.
Preferably, the making the candidate feature object at the current low level be the value object of the candidate feature object at the current high level in the iteration of the next copy operation and feature optimization operation specifically includes:
the candidate feature object at the current low level is made to be a value object of the candidate feature object at the current high level in the iteration of the next copying operation and feature optimization operation according to the preset second probability; wherein the second probability isA i (k) B is the candidate feature object of the current high order i (k) For the candidate feature object at the current low level, f (x) is an affinity function, alpha is a population diversity coefficient, and alpha >0。
Another embodiment of the present invention provides a face region recognition apparatus, including:
the image acquisition module is used for acquiring an image to be identified and randomly selecting a plurality of elliptical areas as initial characteristic objects;
the calculating module is used for calculating the affinity of the initial characteristic object;
the copying module is used for selecting a plurality of initial feature objects to carry out copying operation according to the affinity degree to obtain a plurality of feature objects to be optimized;
the optimization module is used for carrying out feature optimization operation on the feature object to be optimized to obtain a plurality of candidate feature objects;
the iteration module is used for carrying out iteration selection on the candidate feature objects according to a preset selection rule so as to select a part of the candidate feature objects to carry out the next iteration of the copying operation and the feature optimization operation and simultaneously determine the current optimal feature objects;
and the result acquisition module is used for stopping iteration selection when the iteration times reach a preset first threshold value or when the difference value of affinities of two previous and subsequent iterations corresponding to the current optimal feature object is smaller than a preset second threshold value, and taking the current optimal feature object as a face area.
Another embodiment of the present invention correspondingly provides an apparatus for using a face region recognition method, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, where the processor implements the face region recognition method according to any one of the above.
Still another embodiment of the present invention provides a computer readable storage medium, where the computer readable storage medium includes a stored computer program, where the computer program when executed controls a device in which the computer readable storage medium is located to perform a face area identifying method according to any one of the above claims.
Compared with the prior art, the face region identification method, the face region identification device and the storage medium provided by the embodiment of the invention expand the range of the face region candidate solution through the evolutionary selection algorithm, achieve the purpose of accurately and rapidly identifying the face region, and are simple and easy to operate.
Drawings
Fig. 1 is a flow chart of a face region recognition method according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram of a face area recognition device according to an embodiment of the present invention;
Fig. 3 is a schematic diagram of an apparatus using a face region recognition method according to an embodiment of 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.
Referring to fig. 1, a flow chart of a face region recognition method according to an embodiment of the present invention is shown, where the method includes steps S1 to S6:
s1, acquiring an image to be identified, and randomly selecting a plurality of elliptical areas as initial characteristic objects;
s2, calculating the affinity of the initial feature object;
s3, selecting a plurality of initial feature objects to copy according to the affinity degree to obtain a plurality of feature objects to be optimized;
s4, performing feature optimization operation on the feature object to be optimized to obtain a plurality of candidate feature objects;
s5, carrying out iterative selection on the candidate feature objects according to a preset selection rule, and determining a current optimal feature object;
And S6, stopping iteration selection when the iteration times reach a preset first threshold value or when the difference value of affinities of two previous and subsequent iterations corresponding to the current optimal feature object is smaller than a preset second threshold value, and taking the current optimal feature object as a face area.
It should be noted that the face target area has the remarkable characteristics: (1) The color of the face often has a large difference from the background color; (2) there are 5 salient feature points for a face: inner canthus, nasal tip, nasal root, mouth and ear, thus presenting regular facial features in the elliptical area of the face; (3) The color of the face and the background color have jump on the gray value, and the 5 face characteristic points and the face have more uniform gray; (4) The specific size and position of the face in different images are uncertain, but the length axis of the elliptical area is changed to a certain extent, and a maximum and minimum length axis ratio exists. Based on these features, corresponding features can be extracted on the basis of the gray-scale image.
The facial features are main features of the face, and through gray level analysis of scanning in the horizontal direction of the image, the image part of the face can be found to have obvious change features on gray level distribution relative to images of other parts.
The human face region identification is essentially to search the region which is most in line with the human face features in the image space, namely to search the region with the largest human face feature value in the image space. The key to using evolutionary selection algorithms for face region identification is to construct an ideal affinity function so that the function can characterize the face region features. The most obvious characteristic on the gray level image is that the gray level value of the human face area fluctuates frequently and the difference between the high gray level value and the low gray level value is large.
Evolutionary selection algorithms rely on coding to achieve searches independent of the problem itself and represent a better solution potential. In the evolutionary selection algorithm, in the evolutionary process, copying is carried out near each generation of candidate solutions according to the size of affinity, and a group of variant solutions is generated, so that the search range is enlarged (namely, the diversity of feature objects is increased); meanwhile, global searching and local searching are realized, and evolutionary precocity and search trapping in local minima are prevented; meanwhile, the convergence speed is increased through copy selection. The problem of one low dimensional space (N-dimensional) is translated into a higher dimensional (N-dimensional) space to be solved, and then the result is projected into the low dimensional space (N-dimensional), thereby obtaining a more comprehensive understanding of the problem.
The foregoing is considered as illustrative of the principles of the present invention and is provided to facilitate its understanding and to provide a further explanation of the principles of the invention.
Specifically, in the evolutionary selection algorithm, an initial feature object is generally randomly generated, so that an image to be identified is acquired, and a plurality of elliptical areas are randomly selected as the initial feature object, so that an ellipse is selected because the face area is generally elliptical.
The affinity of the initial feature object is calculated. In general, an affinity function is used to evaluate the goodness of each feature object. In the face region recognition problem, whether the region represented by the feature object meets the face features is measured by an affinity function.
And selecting a plurality of initial feature objects according to the affinity to perform copying operation to obtain a plurality of feature objects to be optimized. In general, several initial feature objects with high affinity are selected, because the higher affinity, the greater the likelihood of satisfying the face feature is represented. The copy operation may expand the scope of the search to enable global and local searches to expedite the determination of the face region.
And carrying out feature optimization operation on the feature object to be optimized to obtain a plurality of candidate feature objects. The step is also to expand the search range and quickly find the optimal solution to determine the face area.
And carrying out iterative selection on the candidate feature objects according to a preset selection rule, and determining the current optimal feature object. In each iteration, the generated feature objects always have a score of merit, so that the feature objects are screened to select a better feature object to enter the next iteration.
And stopping iteration selection when the iteration times reach a preset first threshold value or when the difference value of affinities of two previous and subsequent iterations corresponding to the current optimal feature object is smaller than a preset second threshold value, and taking the current optimal feature object as a face area. Generally, before the face region recognition is started, the initialization process needs to set the iteration number, and when the iteration number is reached, the search is ended. When the difference value of affinities of the previous and subsequent iterations corresponding to the current optimal feature object is smaller than a preset second threshold value, the error that the region corresponding to the current optimal feature object is the face region is indicated to be within an allowable range. The second threshold is typically less than 1.
According to the face region identification method provided by the embodiment 1 of the invention, the range of the face region candidate solution is enlarged through the evolution selection algorithm, the purpose of accurately and rapidly identifying the face region is achieved, and the method is simple and easy to operate.
As an improvement of the above solution, the acquiring the image to be identified, randomly selecting a plurality of elliptical areas as the initial feature objects specifically includes:
acquiring an image to be identified, and establishing a coordinate axis by taking the lower left corner of the image to be identified as a circle center; the width of the image to be identified is x, the length of the image to be identified is y, and the sitting mark of the image to be identified is (x, y);
setting the coding of the elliptical area as (i, j, w, h); wherein (i, j) is the lower left angular coordinate of the oval-shaped region circumscribed rectangle, w is the minor axis of the oval-shaped region, and h is the major axis of the oval-shaped region;
randomly selecting a plurality of elliptical areas as initial feature objects by adopting an initialization mode of comprehensive random and priori knowledge according to constraint conditions; wherein, the constraint condition is that i is more than or equal to 0 and less than or equal to x, j is more than or equal to 0 and less than or equal to y, i+w is more than or equal to 0 and less than or equal to x, and j+h is more than or equal to 0 and less than or equal to y.
Specifically, an image to be identified is obtained, and a coordinate axis is established by taking the lower left corner of the image to be identified as the center of a circle; the width of the image to be identified is x, the length of the image to be identified is y, and the sitting mark of the image to be identified is (x, y). The coordinates of the image to be identified are determined in order to facilitate the encoding of the initial feature object.
Setting the coding of the elliptical area as (i, j, w, h); where (i, j) is the lower left angular position of the rectangle circumscribed by the elliptical region, w is the minor axis of the elliptical region, and h is the major axis of the elliptical region. The encoding of the elliptical region is determined and the position of the elliptical region can be known from the encoding.
Because the face positions in many images are relatively centered, a plurality of elliptical areas can be randomly selected as initial feature objects according to constraint conditions by adopting an initialization mode of comprehensive random and priori knowledge; wherein, the constraint condition is that i is more than or equal to 0 and less than or equal to x, j is more than or equal to 0 and less than or equal to y, i+w is more than or equal to 0 and less than or equal to x, and j+h is more than or equal to 0 and less than or equal to y. These constraints are present because the face region cannot be out of the full image. Randomly selecting a number of elliptical areas as initial feature objects can be seen as randomly selecting a number of times, preferably a number of times [10, 50].
As an improvement of the above solution, the calculating the affinity of the initial feature object specifically includes:
calculating the density of the gray scale fluctuation times of the initial characteristic object;
calculating the gray level difference of the peak mean value and the trough mean value of the initial feature object;
and obtaining the affinity of the initial characteristic object according to the density and the gray level difference.
Specifically, the affinity function is used to evaluate the goodness of each feature object. In this problem, it is required to measure whether the region represented by the feature object satisfies the face feature by using an affinity function. According to the analysis of the face characteristics, the statistics of the gray scale fluctuation times and the peak-trough gray scale values of each row are needed in the elliptical area represented by the characteristic object, then,
Calculating the density of the gray scale fluctuation times of the initial characteristic object, and representing the density by D; calculating the gray level difference of the peak mean value and the trough mean value of the initial characteristic object, wherein the gray level difference is expressed by delta g; the larger D and Δg, the greater the likelihood that the region belongs to a face region.
The affinity of the initial feature object can be obtained from the density and gray level differences. In practical application, the ratio of the minor axis to the major axis of the elliptical region, namely w and h, needs to be considered, and if the ratio is in a reasonable range, the function value is unchanged; otherwise a factor less than 1 is required. Thus, the set affinity function can be expressed as:
wherein r is a proportionality coefficient for enhancing the function of the density D in the affinity function value, and generally 0.8 is taken; t is a proportionality coefficient, typically 0.4.
As an improvement of the above solution, the selecting a plurality of initial feature objects according to the affinity to perform the copying operation to obtain a plurality of feature objects to be optimized specifically includes:
the initial feature objects are arranged in a descending order according to the affinity degree, and a plurality of initial feature objects arranged in front are selected;
copying the selected plurality of initial feature objects to obtain a plurality of feature objects to be optimized; wherein the copy operation is C j For the initial feature object, I j Q is an element value of 1 j A dimension vector, j is more than or equal to 1 and less than or equal to k,q j for the initial feature object C j After replicationScale of (2); int []K is the number of the initial feature objects selected as the upper rounding function, N c For the number of the feature objects to be optimized obtained after the copying operation, k is more than or equal to 1 and less than N c ；f(C j ) As a function of affinity.
Specifically, the initial feature objects are arranged in a descending order according to the affinity, and a plurality of initial feature objects arranged in front are selected, in this embodiment, k initial feature objects are selected.
Copying the selected multiple initial feature objects to obtain multiple feature objects to be optimized; wherein the copy operation isC j For initial feature object, I j Q is an element value of 1 j The dimension vector, Θ, here means the operator x; j is more than or equal to 1 and less than or equal to k, and is more than or equal to>q j For initial feature object C j The scale after replication; int []As an upper rounding function, k is the number of selected initial feature objects, N c For the number of feature objects to be optimized obtained after the copying operation, k is more than or equal to 1 and less than N c ；f(C j ) As a function of affinity.
In theory, the feature object to be optimized is identical to the original feature object, and this step is only prepared for subsequent mutation.
As an improvement of the above solution, the performing feature optimization operation on the feature object to be optimized to obtain a plurality of candidate feature objects specifically includes:
selecting one or more feature objects to be optimized according to a preset first probability, and acquiring codes of the feature objects to be optimized;
performing mutation operation on one or more bits of the codes of the feature object to be optimized; wherein the code comprises four bits, i, j, w, h respectively;
and integrating the mutated feature object to be optimized with the non-mutated feature object to be optimized to obtain a plurality of candidate feature objects.
Specifically, selecting one or more feature objects to be optimized according to a preset first probability, and acquiring codes of the feature objects to be optimized;
performing mutation operation on one or more bits of the codes of the feature object to be optimized; wherein the code comprises four bits i, j, w, h respectively. Because the coordinates (i, j) and the minor and major axes w, h are relatively independent, the variant alignment is unaffected. In addition, the optimal feature object is generally located near the feature object with high affinity, and the para-position variation is beneficial to quickly finding the optimal solution. Preferably, mutation operation is randomly carried out on one bit of the codes of the feature object to be optimized.
The mutated feature objects to be optimized and the non-mutated feature objects to be optimized are integrated to obtain a plurality of candidate feature objects, so that more feature objects with more types than the original number can be obtained, the search range is enlarged, and the face area can be found rapidly.
As an improvement of the above solution, the iteratively selecting the candidate feature objects according to a preset selection rule, so as to select a part of the candidate feature objects to perform the next iteration of the copy operation and the feature optimization operation, and simultaneously determine the current optimal feature object, which specifically includes:
sorting the candidate feature objects in a descending order according to the affinity, and selecting a plurality of candidate feature objects arranged in front;
sequentially comparing the affinity of two adjacent candidate feature objects;
if the affinity of the candidate feature object at the current high level is smaller than that of the candidate feature object at the current low level, enabling the candidate feature object at the current low level to be a value object of the candidate feature object at the current high level in the iteration of the next copying operation and feature optimization operation;
if the affinity of the candidate feature object at the current high level is greater than or equal to that of the candidate feature object at the current low level and the affinity of the candidate feature object at the current high level is not the highest, making the candidate feature object at the current low level be a value object of the candidate feature object at the current high level in the iteration of the next copying operation and feature optimizing operation;
If the affinity of the candidate feature object at the current high level is greater than or equal to that of the candidate feature object at the current low level and the affinity of the candidate feature object at the current high level is the highest, making the candidate feature object at the current high level be a value object of the candidate feature object at the current high level in the iteration of the next copying operation and feature optimizing operation;
and after the next iteration value object of each candidate feature object is determined, taking the candidate feature object with the highest affinity as the current optimal feature object.
Specifically, the candidate feature objects are ordered in a descending order according to the affinity, a plurality of candidate feature objects arranged in front are selected, and then the affinity of two adjacent candidate feature objects is sequentially compared;
if the affinity of the current high-order candidate feature object is smaller than that of the current low-order candidate feature object, the current low-order candidate feature object becomes a value object of the current high-order candidate feature object in the iteration of the next copying operation and feature optimization operation, and at this time, the optimal feature object of the population is not obtained yet, and iteration solution needs to be continued.
If the affinity of the current high-order candidate feature object is greater than or equal to that of the current low-order candidate feature object and the affinity of the current high-order candidate feature object is not the highest, the current low-order candidate feature object is made to be a value object of the current high-order candidate feature object in the iteration of the next copying operation and feature optimization operation, so that population diversity is reserved, searching for an optimal solution is continued, and the original solution corresponding to the optimal solution is prevented from being lost due to substitution errors.
If the affinity of the current high-order candidate feature object is greater than or equal to that of the current low-order candidate feature object and the affinity of the current high-order candidate feature object is the highest, the current high-order candidate feature object is made to be a value object of the current high-order candidate feature object in the iteration of the next copying operation and feature optimization operation, and the situation corresponds to the time when the possibility that the current high-order candidate feature object is the optimal feature object is extremely high.
And after the next iteration value object of each candidate feature object is determined, the candidate feature object with the highest affinity is taken as the current optimal feature object.
To facilitate an understanding of the present embodiment, some letters and expressions are provided herein for the description of the present embodiment. Set candidate feature object A i (k)，B i (k) For the corresponding feature object with the maximum affinity after the copy and mutation operation, if f (A i (k))＜f(B i (k) For example), let A i (k+1)＝B i (k) The method comprises the steps of carrying out a first treatment on the surface of the When f (A) i (k))≥f(B i (k) And A) i (k) When the object is not the optimal characteristic object of the current population, let A i (k+1)＝B i (k) The method comprises the steps of carrying out a first treatment on the surface of the When f (A) i (k))≥f(B i (k) And A) i (k) When the object is the optimal characteristic object of the current population, let A i (k+1)＝A i (k)。
As an improvement of the above solution, the making the candidate feature object at the low level of the current be the value object of the candidate feature object at the high level of the current in the iteration of the next copy operation and feature optimization operation specifically includes:
the candidate feature object at the current low level is made to be a value object of the candidate feature object at the current high level in the iteration of the next copying operation and feature optimization operation according to the preset second probability; wherein the second probability isA i (k) B is the candidate feature object of the current high order i (k) For the candidate feature object at the current low level, f (x) is an affinity function, alpha is a population diversity coefficient, and alpha>0。
Specifically, the candidate feature object at the current low level is made into the candidate feature object at the preset second probabilityThe value object of the candidate feature object of the current high level in the iteration of the next copying operation and feature optimization operation; wherein the second probability is A i (k) For the candidate feature object of the current high order, B i (k) For the current low-order candidate feature object, f (x) is an affinity function, alpha is a population diversity coefficient, and alpha>0 is a value related to the population diversity of the candidate feature object. Generally, the better the population diversity, the greater the alpha value and vice versa.
According to the second probability p k Decision assignment A i (k+1) the greater the second probability, the more likely it is to perform A i (k+1)＝B i (k) Thereby increasing the diversity of candidate feature objects and expanding the search range. Determining the assignment based on the second probability is advantageous in avoiding too large or too small a search range.
Referring to fig. 2, a schematic structural diagram of a face area recognition device according to an embodiment of the present invention includes:
the image acquisition module 11 is used for acquiring an image to be identified, and randomly selecting a plurality of elliptical areas as initial characteristic objects;
a calculation module 12, configured to calculate an affinity of the initial feature object;
the replication module 13 is configured to select a plurality of initial feature objects according to the affinity to perform replication operation, so as to obtain a plurality of feature objects to be optimized;
the optimizing module 14 is configured to perform feature optimization operation on the feature object to be optimized to obtain a plurality of candidate feature objects;
The iteration module 15 is configured to perform iterative selection on the candidate feature objects according to a preset selection rule, so as to select a part of the candidate feature objects to perform the next iteration of the copy operation and the feature optimization operation, and determine the current optimal feature object at the same time;
and the result obtaining module 16 is configured to stop the iterative selection when the number of iterations reaches a preset first threshold or when the difference between affinities of two previous and subsequent iterations corresponding to the current optimal feature object is smaller than a preset second threshold, and take the current optimal feature object as a face area.
Preferably, the image acquisition module 11 specifically includes:
the image acquisition unit to be identified is used for acquiring an image to be identified and establishing a coordinate axis by taking the lower left corner of the image to be identified as the center of a circle; the width of the image to be identified is x, the length of the image to be identified is y, and the sitting mark of the image to be identified is (x, y);
a coding unit for setting the coding of the elliptical area to (i, j, w, h); wherein (i, j) is the lower left angular coordinate of the oval-shaped region circumscribed rectangle, w is the minor axis of the oval-shaped region, and h is the major axis of the oval-shaped region;
the selection unit is used for randomly selecting a plurality of elliptical areas as initial characteristic objects by adopting an initialization mode of comprehensive random and priori knowledge according to constraint conditions; wherein, the constraint condition is that i is more than or equal to 0 and less than or equal to x, j is more than or equal to 0 and less than or equal to y, i+w is more than or equal to 0 and less than or equal to x, and j+h is more than or equal to 0 and less than or equal to y.
Preferably, the computing module 12 specifically includes:
a density calculating unit for calculating a density of the number of gray scale undulations of the initial feature object;
the gray level difference calculation unit is used for calculating the gray level difference of the peak mean value and the trough mean value of the initial characteristic object;
and the affinity calculation unit is used for obtaining the affinity of the initial characteristic object according to the density and the gray level difference.
Preferably, the replication module 13 specifically includes:
the first ordering unit is used for arranging the initial feature objects in a descending order according to the affinity degree, and selecting a plurality of initial feature objects arranged in front;
the copying unit is used for copying the selected plurality of initial feature objects to obtain a plurality of feature objects to be optimized; wherein the copy operation isC j For the initial feature object, I j Q is an element value of 1 j Dimension vector, 1.ltoreq.j.ltoreq.k,>q j for the initial feature object C j The scale after replication; int []K is the number of the initial feature objects selected as the upper rounding function, N c For the number of the feature objects to be optimized obtained after the copying operation, k is more than or equal to 1 and less than N c ；f(C j ) As a function of affinity.
Preferably, the optimizing module 14 specifically includes:
The code acquisition unit is used for selecting one or more feature objects to be optimized according to a preset first probability and acquiring codes of the feature objects to be optimized;
the mutation unit is used for performing mutation operation on one or more bits of the codes of the feature object to be optimized; wherein the code comprises four bits, i, j, w, h respectively;
and the integration unit is used for integrating the mutated feature object to be optimized and the non-mutated feature object to be optimized to obtain a plurality of candidate feature objects.
Preferably, the iteration module 15 specifically includes:
the second sorting unit is used for sorting the candidate feature objects in a descending order according to the size of the affinity, and selecting a plurality of candidate feature objects arranged in front;
the comparison unit is used for sequentially comparing the affinities of the two adjacent candidate feature objects;
the first assignment unit is used for enabling the candidate feature object at the current low level to be a value object of the candidate feature object at the current high level in the iteration of the next copying operation and feature optimization operation if the affinity of the candidate feature object at the current high level is smaller than that of the candidate feature object at the current low level;
A second assignment unit, configured to make the candidate feature object at the current low level be a value object of the candidate feature object at the current high level in the iteration of the next copy operation and feature optimization operation if the affinity of the candidate feature object at the current high level is greater than or equal to the affinity of the candidate feature object at the current low level and the affinity of the candidate feature object at the current high level is not the highest;
a third assignment unit, configured to make the current high-level candidate feature object be a value object of the current high-level candidate feature object in the iteration of the next copy operation and feature optimization operation if the affinity of the current high-level candidate feature object is greater than or equal to the affinity of the current low-level candidate feature object and the affinity of the current high-level candidate feature object is the highest;
and the current optimal characteristic object determining unit is used for taking the candidate characteristic object with the highest affinity degree as the current optimal characteristic object after the next iteration value object of each candidate characteristic object is determined.
Preferably, the second assignment unit is specifically configured to:
the candidate feature object at the current low level is made to be a value object of the candidate feature object at the current high level in the iteration of the next copying operation and feature optimization operation according to the preset second probability; wherein the second probability is A i (k) B is the candidate feature object of the current high order i (k) For the currently low order candidate feature object, f (x) is an affinity function, alpha is a value related to the population diversity of the candidate feature object, alpha>0。
The face region recognition device provided by the embodiment of the present invention can implement all the processes of the face region recognition method described in any one of the above embodiments, and the functions and the implemented technical effects of each module and unit in the device are respectively the same as those of the face region recognition method described in the above embodiment, and are not described herein again.
Referring to fig. 3, a schematic diagram of an apparatus for using a face region recognition method according to an embodiment of the present invention includes a processor 10, a memory 20, and a computer program stored in the memory 20 and configured to be executed by the processor 10, where the processor 10 implements the face region recognition method according to any one of the foregoing embodiments when executing the computer program.
By way of example, a computer program may be partitioned into one or more modules/units that are stored in the memory 20 and executed by the processor 10 to perform the present invention. One or more of the modules/units may be a series of computer program instruction segments capable of performing a specific function, the instruction segments describing the execution of a computer program in a face region recognition method. For example, the computer program may be divided into an image acquisition module, a calculation module, a replication module, an optimization module, an iteration module, and a result acquisition module, each of which specifically functions as follows:
The image acquisition module 11 is used for acquiring an image to be identified, and randomly selecting a plurality of elliptical areas as initial characteristic objects;
a calculation module 12, configured to calculate an affinity of the initial feature object;
the replication module 13 is configured to select a plurality of initial feature objects according to the affinity to perform replication operation, so as to obtain a plurality of feature objects to be optimized;
the optimizing module 14 is configured to perform feature optimization operation on the feature object to be optimized to obtain a plurality of candidate feature objects;
the iteration module 15 is configured to perform iterative selection on the candidate feature objects according to a preset selection rule, so as to select a part of the candidate feature objects to perform the next iteration of the copy operation and the feature optimization operation, and determine the current optimal feature object at the same time;
and the result obtaining module 16 is configured to stop the iterative selection when the number of iterations reaches a preset first threshold or when the difference between affinities of two previous and subsequent iterations corresponding to the current optimal feature object is smaller than a preset second threshold, and take the current optimal feature object as a face area.
The device using the face area recognition method can be a computing device such as a desktop computer, a notebook computer, a palm computer, a cloud server and the like. The device using the face region recognition method may include, but is not limited to, a processor, a memory. It will be appreciated by those skilled in the art that the schematic diagram 3 is merely an example of an apparatus using the face region recognition method, and is not limited thereto, and may include more or less components than those illustrated, or may be combined with certain components, or different components, for example, the apparatus using the face region recognition method may further include an input/output device, a network access device, a bus, and the like.
The processor 10 may be a central processing unit (Central Processing Unit, CPU), but may also be other general purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), off-the-shelf programmable gate arrays (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or the like. The general purpose processor may be a microprocessor or the processor 10 may be any conventional processor or the like, and the processor 10 is a control center of the apparatus using the face region recognition method, and connects the respective parts of the entire apparatus using the face region recognition method using various interfaces and lines.
The memory 20 may be used to store the computer program and/or module, and the processor 10 implements various functions of the apparatus using the face region recognition method by running or executing the computer program and/or module stored in the memory 20 and invoking data stored in the memory 20. The memory 20 may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, application programs required for at least one function, and the like; the storage data area may store data created according to program use, or the like. In addition, the memory 20 may include high-speed random access memory, and may also include nonvolatile memory, such as a hard disk, memory, plug-in hard disk, smart Media Card (SMC), secure Digital (SD) Card, flash Card (Flash Card), at least one disk storage device, flash memory device, or other volatile solid state storage device.
Wherein the module integrated with the apparatus using the face region recognition method may be stored in a computer readable storage medium if implemented in the form of a software functional unit and sold or used as a separate product. Based on such understanding, the present invention may implement all or part of the flow of the method of the above embodiment, or may be implemented by a computer program to instruct related hardware, where the computer program may be stored in a computer readable storage medium, and when the computer program is executed by a processor, the steps of each method embodiment may be implemented. The computer program comprises computer program code, and the computer program code can be in a source code form, an object code form, an executable file or some intermediate form and the like. The computer readable medium may include: any entity or device capable of carrying computer program code, a recording medium, a U disk, a removable hard disk, a magnetic disk, an optical disk, a computer Memory, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), an electrical carrier signal, a telecommunications signal, a software distribution medium, and so forth. It should be noted that the content of the computer readable medium can be appropriately increased or decreased according to the requirements of the jurisdiction's jurisdiction and the patent practice, for example, in some jurisdictions, the computer readable medium does not include electrical carrier signals and telecommunication signals according to the jurisdiction and the patent practice.
The embodiment of the invention also provides a computer readable storage medium, which comprises a stored computer program, wherein when the computer program runs, the equipment where the computer readable storage medium is located is controlled to execute the face area identification method according to any one of the above embodiments.
In summary, the face region recognition method, the device and the storage medium provided by the embodiment of the invention expand the range of face region candidate solutions through the evolution selection algorithm, achieve the purpose of accurately and rapidly recognizing the face region, and can accurately recognize the face region even when the illumination condition is insufficient or the light condition is dark.
While the foregoing is directed to the preferred embodiments of the present invention, it will be appreciated by those skilled in the art that changes and modifications may be made without departing from the principles of the invention, such changes and modifications are also intended to be within the scope of the invention.
Claims (9)
1. The face area recognition method is characterized by comprising the following steps of:
acquiring an image to be identified, and randomly selecting a plurality of elliptic areas as initial characteristic objects;
calculating the affinity of the initial feature object, wherein the calculating the affinity of the initial feature object specifically comprises: calculating the density of the gray scale fluctuation times of the initial feature object, calculating the gray scale difference between the peak mean value and the trough mean value of the initial feature object, and obtaining the affinity of the initial feature object according to the density and the gray scale difference;
selecting a plurality of initial feature objects to copy according to the affinity degree to obtain a plurality of feature objects to be optimized;
performing feature optimization operation on the feature object to be optimized to obtain a plurality of candidate feature objects;
performing iterative selection on the candidate feature objects according to a preset selection rule, so as to select a part of the candidate feature objects to perform the next iteration of the copying operation and the feature optimization operation, and simultaneously determining the current optimal feature objects;
and stopping iteration selection when the iteration times reach a preset first threshold or when the difference value of affinities of two previous and subsequent iterations corresponding to the current optimal feature object is smaller than a preset second threshold, and taking the current optimal feature object as a face area.
2. The face region recognition method according to claim 1, wherein the acquiring the image to be recognized, randomly selecting a plurality of elliptical regions as initial feature objects, comprises:
acquiring an image to be identified, and establishing a coordinate axis by taking the lower left corner of the image to be identified as a circle center; the width of the image to be identified is x, the length of the image to be identified is y, and the sitting mark of the image to be identified is (x, y);
setting the coding of the elliptical area as (i, j, w, h); wherein (i, j) is the lower left angular coordinate of the oval-shaped region circumscribed rectangle, w is the minor axis of the oval-shaped region, and h is the major axis of the oval-shaped region;
randomly selecting a plurality of elliptical areas as initial feature objects by adopting an initialization mode of comprehensive random and priori knowledge according to constraint conditions; wherein, the constraint condition is that i is more than or equal to 0 and less than or equal to x, j is more than or equal to 0 and less than or equal to y, i+w is more than or equal to 0 and less than or equal to x, and j+h is more than or equal to 0 and less than or equal to y.
3. The face region recognition method according to claim 1, wherein the selecting a plurality of initial feature objects according to the affinity to perform the copying operation to obtain a plurality of feature objects to be optimized specifically includes:
the initial feature objects are arranged in a descending order according to the affinity degree, and a plurality of initial feature objects arranged in front are selected;
Copying the selected plurality of initial feature objects to obtain a plurality of feature objects to be optimized; wherein the copy operation isC j For the initial feature object, I j Q is an element value of 1 j A dimension vector, j is more than or equal to 1 and less than or equal to k,1≤s≤j,q j for the initial feature object C j The scale after replication; int []K is the number of the initial feature objects selected as the upper rounding function, N c For the number of the feature objects to be optimized obtained after the copying operation, k is more than or equal to 1 and less than N c ；f(C j ) As a function of affinity.
4. The face region recognition method according to claim 2, wherein the feature optimization operation is performed on the feature object to be optimized to obtain a plurality of candidate feature objects, and the method specifically includes:
selecting one or more feature objects to be optimized according to a preset first probability, and acquiring codes of the feature objects to be optimized;
performing mutation operation on one or more bits of the codes of the feature object to be optimized; wherein the code comprises four bits, i, j, w, h respectively;
and integrating the mutated feature object to be optimized with the non-mutated feature object to be optimized to obtain a plurality of candidate feature objects.
5. A face region recognition method according to claim 3, wherein the iterative selection of the candidate feature objects according to a preset selection rule is performed to select a part of the candidate feature objects for performing the next iteration of the copying operation and the feature optimization operation, and the method further comprises determining the current optimal feature object at the same time, and specifically includes:
Sorting the candidate feature objects in a descending order according to the affinity, and selecting a plurality of candidate feature objects arranged in front;
sequentially comparing the affinity of two adjacent candidate feature objects;
if the affinity of the candidate feature object at the current high level is smaller than that of the candidate feature object at the current low level, enabling the candidate feature object at the current low level to be a value object of the candidate feature object at the current high level in the iteration of the next copying operation and feature optimization operation;
if the affinity of the candidate feature object at the current high level is greater than or equal to that of the candidate feature object at the current low level and the affinity of the candidate feature object at the current high level is not the highest, making the candidate feature object at the current low level be a value object of the candidate feature object at the current high level in the iteration of the next copying operation and feature optimizing operation;
if the affinity of the candidate feature object at the current high level is greater than or equal to that of the candidate feature object at the current low level and the affinity of the candidate feature object at the current high level is the highest, making the candidate feature object at the current high level be a value object of the candidate feature object at the current high level in the iteration of the next copying operation and feature optimizing operation;
And after the next iteration value object of each candidate feature object is determined, taking the candidate feature object with the highest affinity as the current optimal feature object.
6. The face region recognition method according to claim 5, wherein the making the candidate feature object at the low level the candidate feature object at the high level is a value object in an iteration of a next copy operation and a feature optimization operation, specifically includes:
the candidate feature object at the current low level is made to be a value object of the candidate feature object at the current high level in the iteration of the next copying operation and feature optimization operation according to the preset second probability; wherein the second probability isA i (k) B is the candidate feature object of the current high order i (k) For the candidate feature object at the current low level, alpha is a population diversity coefficient, alpha>0。
7. A face region recognition apparatus, comprising:
the image acquisition module is used for acquiring an image to be identified and randomly selecting a plurality of elliptical areas as initial characteristic objects;
the calculating module is configured to calculate an affinity of the initial feature object, where the calculating the affinity of the initial feature object specifically includes: calculating the density of the gray scale fluctuation times of the initial feature object, calculating the gray scale difference between the peak mean value and the trough mean value of the initial feature object, and obtaining the affinity of the initial feature object according to the density and the gray scale difference;
The copying module is used for selecting a plurality of initial feature objects to carry out copying operation according to the affinity degree to obtain a plurality of feature objects to be optimized;
the optimization module is used for carrying out feature optimization operation on the feature object to be optimized to obtain a plurality of candidate feature objects;
the iteration module is used for carrying out iteration selection on the candidate feature objects according to a preset selection rule so as to select a part of the candidate feature objects to carry out the next iteration of the copying operation and the feature optimization operation and simultaneously determine the current optimal feature objects;
and the result acquisition module is used for stopping iteration selection when the iteration times reach a preset first threshold value or when the difference value of affinities of two previous and subsequent iterations corresponding to the current optimal feature object is smaller than a preset second threshold value, and taking the current optimal feature object as a face area.
8. An apparatus using a face region recognition method, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, the processor implementing the face region recognition method according to any one of claims 1 to 6 when executing the computer program.
9. A computer readable storage medium, characterized in that the computer readable storage medium comprises a stored computer program, wherein the computer program, when run, controls a device in which the computer readable storage medium is located to perform the face area recognition method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010050220.0A CN111259806B (en) | 2020-01-16 | 2020-01-16 | Face area identification method, device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010050220.0A CN111259806B (en) | 2020-01-16 | 2020-01-16 | Face area identification method, device and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111259806A CN111259806A (en) | 2020-06-09 |
| CN111259806B true CN111259806B (en) | 2023-11-14 |
Family
ID=70950597
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010050220.0A Active CN111259806B (en) | 2020-01-16 | 2020-01-16 | Face area identification method, device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111259806B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112329715A (en) * | 2020-11-25 | 2021-02-05 | 广州杰赛科技股份有限公司 | Face recognition method, device, equipment and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101968853A (en) * | 2010-10-15 | 2011-02-09 | 吉林大学 | Improved immune algorithm based expression recognition method for optimizing support vector machine parameters |
| CN102024141A (en) * | 2010-06-29 | 2011-04-20 | 上海大学 | Face recognition method based on Gabor wavelet transform and local binary pattern (LBP) optimization |
| CN104268531A (en) * | 2014-09-30 | 2015-01-07 | 江苏中佑石油机械科技有限责任公司 | Face feature data obtaining system |
| CN106981066A (en) * | 2017-03-06 | 2017-07-25 | 武汉嫦娥医学抗衰机器人股份有限公司 | A kind of interior face image dividing method based on the colour of skin |
| CN110321217A (en) * | 2019-03-25 | 2019-10-11 | 深圳大学 | A kind of cloud resource dispatching method, device, equipment and the storage medium of multiple target |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7139738B2 (en) * | 2002-06-27 | 2006-11-21 | Koninklijke Philips Electronics N.V. | Face recognition using evolutionary algorithms |
| WO2011135410A1 (en) * | 2010-04-28 | 2011-11-03 | Indian Statistical Institute | Optimization technique using evolutionary algorithms |
| US10902243B2 (en) * | 2016-10-25 | 2021-01-26 | Deep North, Inc. | Vision based target tracking that distinguishes facial feature targets |
-
2020
- 2020-01-16 CN CN202010050220.0A patent/CN111259806B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102024141A (en) * | 2010-06-29 | 2011-04-20 | 上海大学 | Face recognition method based on Gabor wavelet transform and local binary pattern (LBP) optimization |
| CN101968853A (en) * | 2010-10-15 | 2011-02-09 | 吉林大学 | Improved immune algorithm based expression recognition method for optimizing support vector machine parameters |
| CN104268531A (en) * | 2014-09-30 | 2015-01-07 | 江苏中佑石油机械科技有限责任公司 | Face feature data obtaining system |
| CN106981066A (en) * | 2017-03-06 | 2017-07-25 | 武汉嫦娥医学抗衰机器人股份有限公司 | A kind of interior face image dividing method based on the colour of skin |
| CN110321217A (en) * | 2019-03-25 | 2019-10-11 | 深圳大学 | A kind of cloud resource dispatching method, device, equipment and the storage medium of multiple target |
Non-Patent Citations (1)
| Title |
|---|
| 王帆等."基于克隆选择算法的人脸特征选择".《应用科技》.2009,第第36卷卷(第第3期期),第12页右栏第23行-第13页右栏第7行及图1. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111259806A (en) | 2020-06-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107944020B (en) | Face image searching method and device, computer device and storage medium | |
| JP5506785B2 (en) | Fingerprint representation using gradient histogram | |
| US10776400B2 (en) | Clustering using locality-sensitive hashing with improved cost model | |
| CN108334805B (en) | Method and device for detecting document reading sequence | |
| US10438083B1 (en) | Method and system for processing candidate strings generated by an optical character recognition process | |
| CN108491850B (en) | Automatic feature point extraction method and device of three-dimensional tooth mesh model | |
| US7643660B1 (en) | System, method and computer program product for fingerprint verification | |
| CN111476270A (en) | Course information determining method, device, equipment and storage medium based on K-means algorithm | |
| CN112597978B (en) | Fingerprint matching method and device, electronic equipment and storage medium | |
| CN110796135A (en) | Target positioning method and device, computer equipment and computer storage medium | |
| CN111311593B (en) | Multi-ellipse detection and evaluation algorithm, device and terminal | |
| CN112364881B (en) | Advanced sampling consistency image matching method | |
| CN111259806B (en) | Face area identification method, device and storage medium | |
| CN114417095A (en) | Data set partitioning method and device | |
| US6052483A (en) | Methods and apparatus for classification of images using distribution maps | |
| CN112465050A (en) | Image template selection method, device, equipment and storage medium | |
| CN111222558A (en) | Image processing method and storage medium | |
| CN111382760A (en) | Image category identification method and device and computer readable storage medium | |
| CN112329715A (en) | Face recognition method, device, equipment and storage medium | |
| CN112446428B (en) | Image data processing method and device | |
| Kotyza et al. | Detection of directions in an image as a method for circle detection | |
| CN114723791A (en) | Feature point matching method and device, electronic equipment and storage medium | |
| CN109871867A (en) | A kind of pattern fitting method of the data characterization based on preference statistics | |
| CN113900889B (en) | Method and system for intelligently identifying APP manual operation | |
| CN110968690A (en) | Clustering division method and device for words, equipment and storage medium |
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 |