s400, pair C_hk Processing to obtain C_h Corresponding set of target image vector types C'_h ＝{C'_h1 ，……，C'_hβ ，……，C'_ha }，C'_hβ ＝{C'¹_hβ ，……，C'^u_hβ ，……，C'^w(β)_hβ }，C'^u_hβ For the nth target image vector in the nth class in the nth second image data processing platform, u= … … w (β), w (β) is the number of target image vectors in the nth class, β= … … a, a is C'_h The number of target image vector types.

S500, according to A and C'^u_hβ Acquiring a first target image feature set AC= { AC corresponding to A₁ ,……，AC_i ,……，AC_λ }，AC_i Is A_i A corresponding first list of target image features.

S600 according to AC_i The first image data processing platform acquires a second target image feature set AD= { AD corresponding to the A₁ ,……，AD_i ,……，AD_λ }，AD_i Is A_i And a corresponding second target image feature list.

S700, AD_i And sending the message to the ith first target user.

Compared with the prior art, the data processing system for acquiring the facial image features has obvious beneficial effects, can achieve quite technical progress and practicality, has wide industrial utilization value, and has at least the following beneficial effects:

the invention provides a data processing system for acquiring facial image characteristics, which comprises: the system comprises a database, a processor, a memory storing a computer program, a first image data processing platform and N second image data processing platforms, wherein the database comprises a first face image list and a preset image information set, and when the computer program is executed by the processor, the following steps are realized: acquiring a first face image list and a preset image information set from a database, inputting the preset image information set into N second image data processing platforms to acquire a target image vector list corresponding to each second image data processing platform, carrying out clustering processing on target image vectors in each second image data processing platform to acquire a target image vector type set corresponding to each second image data processing platform, acquiring a first target image feature set corresponding to a first face image according to the first face image list and the target image vectors, acquiring a second target image feature set corresponding to the first face image according to the first target image feature and the first image data processing platform, and sending the second target image feature set corresponding to the first face image to a first target user. As can be seen, in one aspect of the invention, the face images in the database are not subjected to one-to-one comparison, the preset image information in the database is subjected to clustering treatment, the comparison is firstly performed with the preset image information type, and then the comparison is performed with the preset image information meeting the preset type, so that the face image searching and identifying range is reduced, and the real-time performance of face identification is improved; on the other hand, the plurality of data query platforms are associated, face images are processed at the same time, the clustering quantity of data in the data query platforms is dynamically adjusted, the data storage space is improved, and the accuracy of the obtained face image features is higher under the condition that information in a database is not lost.

The foregoing description is only an overview of the present invention, and is intended to be implemented in accordance with the teachings of the present invention, as well as the preferred embodiments thereof, together with the following detailed description of the invention, given by way of illustration only, together with the accompanying drawings.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and 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 execution computer program of a data processing system for acquiring facial image features 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 fall within the scope of the invention.

It should be noted that the terms "first," "second," and the like in the description and the claims of the present invention and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the invention described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or server that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed or inherent to such process, method, article, or apparatus, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

A data processing system for acquiring facial image features, the system comprising: the image processing system comprises a database, a processor, a memory storing a computer program, a first image data processing platform and N second image data processing platforms, wherein the database comprises a first face image list and a preset image information set, and when the computer program is executed by the processor, the following steps are realized, as shown in fig. 1:

Specifically, the database further includes an initial face image set of the target user, and when the computer program is executed by the processor, the first face image list is obtained through the following steps:

s110, acquiring a first face image list A and a second face image list B according to the initial face image set of the target user.

Specifically, in S110, the following steps are further included:

s111, acquiring an initial face image set Q= { Q of a target user from a database₁ ，……，Q_g ，……，Q_n }，Q_g For the g-th face image of the target user, g= … … n, n is the number of target users.

S112, according to Q, obtaining an initial time list T= { T corresponding to Q₁ ，……，T_g ，……，T_n }，T_g Is Q_g Corresponding initial time nodes.

S113, when T⁰ -T_g When T 'is less than or equal to T', T is taken as_g Corresponding Q_g Inserting a first initial face image list A '= { A'₁ ，……，A'_q ，……，A'_b }，A'_q For the first initial face image corresponding to the q-th first initial target user, q= … … b, where b is the number of first initial target users, where T⁰ And T' is a preset first time threshold value for the current time node corresponding to the Q.

Specifically, the value range of T 'is 90 to 180 days, and those skilled in the art know that T' is selected according to actual requirements, which is not described herein.

S114, when T⁰ -T_g At > T', T is taken_g Corresponding Q_g Inserting a second initial face image list B '= { B'₁ ，……，B'_L ，……，B'_t }，B'_L For the second initial face image corresponding to the L-th second initial target user, l= … … t, where t is the number of second initial target users.

S115, when b > lambda₀ When a=a 'and b=b' are acquired.

S116, when b is less than or equal to lambda₀ When T 'is replaced by T'₁ Wherein lambda is₀ For a preset image quantity threshold value, T'₁ Is a preset second time threshold.

Specifically, λ is performed according to the data amount storable by the second image data processing platform₀ Is known to those skilled in the art, and is not described in detail herein.

Specifically, T'₁ ＞T'。

By expanding the time range, more initial face images are distributed in the first initial face image set, the comparison quantity of the initial face images with preset image information in the database can be reduced, the face image searching and identifying range is reduced, and the real-time performance of face identification is improved.

S107, when T⁰ -T_L ≤T'₁ When B 'is to be used'_L Inserted into a 'to obtain a=a', wherein T_L Is B'_L Corresponding time nodes.

S108, when T⁰ -T_L ＞T'₁ When B 'is to be used'_j Deleted in B 'to obtain b=b'.

According to the method, the initial face images of the target users are divided and classified according to different times, the time is dynamically adjusted according to the data quantity which can be stored by the second image data processing platform, more face images are input into the second data platform, the face image searching and identifying range can be shortened, and the real-time performance of face identification is improved.

according to the method, the preset image information is input to the N second image data processing platforms, a certain amount of preset image information is stored in each second image data processing platform, and the association relation is established by storing the data information in different platforms, so that the data storage space is reduced, and the accuracy of the obtained face image features is higher under the condition that the database information is not lost.

Specifically, in S400, the following steps are further included:

s401, from C_h Mid-image vector type set H_h ＝{H_h1 ，……，H_hβ ，……，H_ha Sum H_h Corresponding initial center point set H¹_h ＝{H¹_h1 ，……，H¹_hβ ，……，H¹_ha }，H_hβ For the intermediate image vector of the beta type in the H second image data processing platform, H¹_hβ From C when initially unblustered_h The vector of the beta sample obtained in (1), wherein, the initial H_hβ ＝Null。

S402, according to C_h And H¹_h Obtaining C_hk Corresponding target similarity list HC_hk ＝{HC¹_hk ，……，HC^β_hk ，……，HC^a_hk }，HC^β_hk Is C_hk And H is¹_hβ The target similarity between the two can be calculated according to any model by a person skilled in the art, and will not be described here.

S403, when HC^β_hk Is HC (HC)_hk At the minimum of (C)_hk Inserted into H_hβ In order to obtain C'_hβ ＝H_hβ 。

S404, according to H_hβ Obtaining H_h Corresponding intermediate center point set H²_h ＝{H²_h1 ，……，H²_hβ ，……，H²_ha }, wherein H²_hβ Meets the following conditions:

wherein Y is¹_y To H after 1 st clustering_hβ The y-th preset image information, eta₁ (H_hβ ) To H after 1 st clustering_hβ The number of image information is preset.

S405, repeatedly executing S402-S404 to obtain H_h Corresponding target center point set H^E_h ＝{H^E_h1 ，……，H^E_hβ ，……，H^E_ha }, wherein H^E_hβ Is the beta target center point obtained after the clustering of the (E-1) th time.

Specifically, the H^E_hβ Acquisition mode and H of (2)²_hβ The acquisition modes of the obtained images are consistent.

S406, when H^E_hβ ＝H^E-1_hβ At the time, C'_hβ ＝H_hβ 。

According to the method, the preset image information in each second image data processing platform is clustered, similar images are gathered into one type, and when the first face image is compared with the preset image information in the second image data processing platform, the face image recognition search range can be narrowed, so that the real-time performance of face recognition is improved.

Specifically, in S400, a meets the following condition:

a＝r₁ (μ+1)³ +r₂ (μ+1)² +r₃ (μ+1)+r₄ wherein r is₁ For a preset first parameter, r₂ For a preset second parameter, r₃ R is a preset third parameter₄ Is a preset fourth parameter.

Specifically, r₁ The range of the value of (C) is-8 e^-13 ～-5e^-12 Wherein e is the euler constant.

Preferably, r₁ The value of (2) is-4 e^-12 。

Specifically, r₂ The value range of (2) is 2e^-8 ～3e^-6 。

Preferably, r₂ Has a value of 2e^-7 。

Specifically, r₃ The range of the value of (2) is-1 to 1.

Preferably, r₃ Has a value of-0.0068.

Specifically, r₄ The range of the value of (2) is 50-100.

Preferably, r₄ Is 99.025.

Above-mentioned, carry out the adjustment of face image type quantity according to the size of face image data volume, the balance of performance and recall rate when can guaranteeing the data match for when obtaining initial face image feature, guarantee the efficiency of data operation and the comprehensiveness of data match, make the degree of accuracy of the face image feature who obtains higher.

Specifically, in S500, the following steps are further included:

s501 according to A_i And C'_h Obtaining A_i Corresponding first similarity list Fⁱ_h ＝{Fⁱ_h1 ，……，Fⁱ_hβ ，……，Fⁱ_ha }，Fⁱ_hβ Is A_i And C'_hβ Is a first similarity of (c).

Specifically, in S501, the following steps are further included:

s5011, according to A, obtaining a first face image vector list A corresponding to A⁰ ＝{A⁰₁ ，……，A⁰_i ，……，A⁰_λ }，A⁰_i Is A_i A corresponding first face image vector.

Specifically, the model for converting the first face image into the first face image vector is selected by a person skilled in the art according to actual requirements, and will not be described herein.

S5013, according to A⁰ Obtaining A⁰_i ＝(A⁰_i1 ，……，A⁰_ix ，……，A⁰_ip ) Wherein A is⁰_ix Is A⁰_i The bit value of the x-th bit in (1), x= … … p, p is the dimension of the first face image vector.

S5015, according to C'_h Obtaining C'^u_hβ ＝(C'^u1_hβ ，……，C'^ux_hβ ，……，C'^up_hβ ) Wherein C'^ux_hβ Is C'^u_hβ Bit value of the x-th bit in (b).

S5017 according to A⁰_ix And C'^ux_hβ Obtaining a first similarity Fⁱ_hβ Wherein F isⁱ_hβ Meets the following conditions:

s503, when Fⁱ_hβ ≥F₀ At the time, obtain A_i Corresponding second similarity list Fⁱ_hβ ＝{Fⁱ¹_hβ ，……，F^iu_hβ ，……，F^iw(β)_hβ }，F^iu_hβ Is A_i And C'^u_hβ Is of the second similarity of F₀ Is a preset first similarity threshold.

Specifically, F^iu_hβ Acquisition mode and F of (2)ⁱ_hβ The acquisition modes of the obtained images are consistent.

Specifically, F₀ The value range of (2) is 0.85-0.9, and the preset similarity threshold is set by the person skilled in the art according to the actual requirement, and will not be described herein.

S505, when F^iu_hβ ≥F₁ At the time, the AC is acquired_i Wherein F₁ Is a preset second similarity threshold.

Specifically, F₁ The value range of (2) is 0.85-0.9, and the preset similarity threshold is set by the person skilled in the art according to the actual requirement, and will not be described herein.

Specifically, F₀ ＝F₁ 。

Specifically, the second image data processing platform further includes C_h Corresponding first image feature list FC_h ＝{FC_h0 ,……，FC_hk ，……，FC_hμ }，FC_hk ＝{FC¹_hk ，……，FC^θ_hk ，……，FC^ζ_hk }，FC^θ_hk Is C_hk Corresponding to the first record included in the θ -th first field in the first image feature list, θ= … … ζ, where ζ is the number of first fields.

Specifically, in S505, the following steps are further included:

s501, when F^iu_hβ ≥F₁ At the time, F is acquired^iu_hβ Corresponding C'^u_hβ 。

S503, when C'^u_hβ ＝C_hk At the time, the AC is acquired_i ＝FC_hk 。

The first face image is compared with the target image vector types in each second image data processing platform, the target image vector types meeting the conditions are selected according to the preset similarity threshold, and then the comparison is carried out with the target image vectors meeting the conditions, so that the face image searching range can be narrowed, the real-time performance of face recognition is improved, meanwhile, the data storage space is reduced, and the accuracy of the obtained face image features is higher under the condition that database information is not lost.

Specifically, the first image data processing platform further includes a second image feature list M ' = { M ' corresponding to M '₁ ，……，M'_s ，……，M'_v }，M'_s ＝{M'_s1 ，……，M'_sε ，……，M'_sη }，M'_sε Is M_s And the epsilon=1 … … eta, eta is the number of the second fields, of the second records contained in the epsilon second field in the corresponding second image feature list.

Specifically, the step S600 includes the following steps:

s601, when the theta first field is a preset field and the epsilon second field is a preset field, obtaining the FC^θ_hk And M'_sε Wherein the preset field is an ID.

S603, when s=h+k×n, acquiring AD_i ＝M'_s 。

By associating the data in the first image data processing platform with the data in the second image data processing platform, the second image data processing platform with less stored data is subjected to the same ID matching with the first image data processing platform, so that all the information of the first face image is obtained, the searching range of the face image is reduced, the real-time performance of face recognition is improved, and meanwhile, the accuracy of the obtained face image features is higher under the condition that the database information is not lost.

S700, AD_i And sending the message to the ith first target user.

Specifically, the method further comprises the following steps after S700:

s800, obtaining a second face image list B= { B from the database₁ ，……，B_j ，……，B_γ }，B_j For the second face image corresponding to the j-th second target user, j= … … γ, and γ is the number of second target users.

S900, B and M are input into the first image data processing platform, and a second target image feature set BD= { BD corresponding to B is obtained₁ ,……，BD_j ,……，BD_γ }，BD_j Is B_j And a corresponding second target image feature list.

Specifically, in S900, the following steps are further included:

s901, according to B, obtaining a second face image vector list B corresponding to B⁰ ＝{B⁰₁ ，……，B⁰_j ，……，B⁰_γ }，B⁰_j And the j second face image vector is the j second face image vector in the B.

Specifically, the mode of acquiring the second face image vector is consistent with the mode of acquiring the first face image vector, and the fact that the acquired vector forms are inconsistent due to different modes can be avoided by ensuring the consistency of the acquisition modes, the accuracy of subsequent similarity calculation is affected, and therefore the accuracy of the acquired face image features is affected.

S903, obtaining B from B and M_j Corresponding third similarity list BM_j ＝{BM¹_j ，……，BM^s_j ，……，BM^v_j }，BM^s_j Is B_j And M is as follows_s Is a third similarity of (3).

Specifically, BM^s_j Acquisition mode of F^iu_hβ Acquisition mode and BM of (a)^s_j The acquisition modes of the obtained images are consistent.

S905, as BM^s_j ≥F₂ At the time of acquiring BD_j Wherein F is₂ Is a preset third similarity threshold.

Specifically, F₂ The value range of (2) is 0.85-0.9, and the preset similarity threshold is set by the person skilled in the art according to the actual requirement, and will not be described herein.

Specifically, F₀ ＝F₁ ＝F₂ 。

Specifically, in S905, the following steps are further included:

s9051, as BM^s_j ≥F₂ At the time, obtain B_j Corresponding M_s 。

S9053 according to M_s And M'_s Obtaining BD_j ＝{M'_s1 ，……，M'_sε ，……，M'_sη }。

S1000, BD is processed_j And transmitting to the j second target user.

While certain specific embodiments of the invention have been described in detail by way of example, it will be appreciated by those skilled in the art that the above examples are for illustration only and are not intended to limit the scope of the invention. Those skilled in the art will also appreciate that many modifications may be made to the embodiments without departing from the scope and spirit of the invention. The scope of the invention is defined by the appended claims.

Claims

1. A data processing system for acquiring facial image features, the system comprising: the system comprises a database, a processor, a memory storing a computer program, a first image data processing platform and N second image data processing platforms, wherein the database comprises a first face image list and a preset image information set, and when the computer program is executed by the processor, the following steps are realized:

s100, acquiring a first face image list A= { A from a database₁ ，……，A_i ，……，A_λ }，A_i For the first face image corresponding to the ith first target user, i= … … λ, where λ is the number of first target users;

s200, acquiring a preset image information set M= { M from a database₁ ，……，M_s ，……，M_v }，M_s S= … … v for the s-th preset image information in the database, v being the number of preset image information;

s300, inputting M into N second image data processing platforms, and obtaining a target image corresponding to each second image data processing platformVector list C_h ＝{C_h0 ，……，C_hk ，……，C_hμ }, wherein C_hk For the kth target image vector corresponding to the (h) th second image data processing platform, h= … … N, C_hk ＝M_h+k*N ，M_h+k*N For the (h+k x N) th preset image information in M, k=0 … … μ,

s400, pair C_hk Processing to obtain C_h Corresponding set of target image vector types C'_h ＝{C'_h1 ，……，C'_hβ ，……，C'_ha }，C'_hβ ＝{C'¹_hβ ，……，C'^u_hβ ，……，C'^w(β)_hβ }，C'^u_hβ For the nth target image vector in the nth class in the nth second image data processing platform, u= … … w (β), w (β) is the number of target image vectors in the nth class, β= … … a, a is C'_h The number of target image vector types; wherein, in S400, the following steps are further included:

s401, from C_h Mid-image vector type set H_h ＝{H_h1 ，……，H_hβ ，……，H_ha Sum H_h Corresponding initial center point set H¹_h ＝{H¹_h1 ，……，H¹_hβ ，……，H¹_ha }，H_hβ For the intermediate image vector of the beta type in the H second image data processing platform, H¹_hβ From C when initially unblustered_h The vector of the beta sample obtained in (1), wherein, the initial H_hβ ＝Null；

S402, according to C_h And H¹_h Obtaining C_hk Corresponding target similarity list HC_hk ＝{HC¹_hk ，……，HC^β_hk ，……，HC^a_hk }，HC^β_hk Is C_hk And H is¹_hβ Target similarity between the two;

s403, when HC^β_hk Is HC (HC)_hk At the minimum of (C)_hk Inserted into H_hβ In order to obtain C'_hβ ＝H_hβ ；

wherein Y is¹_y To H after 1 st clustering_hβ The y-th preset image information, eta₁ (H_hβ ) To H after 1 st clustering_hβ The number of the preset image information;

s405, repeatedly executing S402-S404 to obtain H_h Corresponding target center point set H^E_h ＝{H^E_h1 ，……，H^E_hβ ，……，H^E_ha }, wherein H^E_hβ The beta target center point is obtained after clustering for the (E-1) th time;

s406, when H^E_hβ ＝H^E-1_hβ At the time, C'_hβ ＝H_hβ ；

S500, according to A and C'^u_hβ Acquiring a first target image feature set AC= { AC corresponding to A₁ ,……，AC_i ,……，AC_λ }，AC_i Is A_i A corresponding first target image feature list; wherein, in S500, the method further comprises the following steps:

s501 according to A_i And C'_h Obtaining A_i Corresponding first similarity list Fⁱ_h ＝{Fⁱ_h1 ，……，Fⁱ_hβ ，……，Fⁱ_ha }，Fⁱ_hβ Is A_i And C'_hβ Wherein in S501 the steps of:

s5011, according to A, obtaining a first face image vector list A corresponding to A⁰ ＝{A⁰₁ ，……，A⁰_i ，……，A⁰_λ }，A⁰_i Is A_i A corresponding first face image vector;

s5013, according to A⁰ Obtaining A⁰_i ＝(A⁰_i1 ，……，A⁰_ix ，……，A⁰_ip ) Wherein A is⁰_ix Is A⁰_i The bit value of the x-th bit in the table, x= … … p, p is the dimension of the first face image vector;

s5015, according to C'_h Obtaining C'^u_hβ ＝(C'^u1_hβ ，……，C'^ux_hβ ，……，C'^up_hβ ) Wherein C'^ux_hβ Is C'^u_hβ Bit value of the x-th bit of the middle;

s503, when Fⁱ_hβ ≥F₀ At the time, obtain A_i Corresponding second similarity list Fⁱ_hβ ＝{Fⁱ¹_hβ ，……，F^iu_hβ ，……，F^iw^(β)_hβ }，F^iu_hβ Is A_i And C'^u_hβ Is of the second similarity of F₀ Is a preset first similarity threshold value, wherein F^iu_hβ Acquisition mode and F of (2)ⁱ_hβ The acquisition modes of the two are consistent;

s505, when F^iu_hβ ≥F₀ At the time, the AC is acquired_i ；

S600 according to AC_i The first image data processing platform acquires a second target image feature set AD= { AD corresponding to the A₁ ,……，AD_i ,……，AD_λ }，AD_i Is A_i A corresponding second target image feature list;

s700, AD_i And sending the message to the ith first target user.

2. The data processing system for acquiring facial image features according to claim 1, wherein in S400, a meets the following condition:

3. The data processing system for acquiring facial image features according to claim 1, wherein said second image data processing platform further comprises C_h Corresponding first image feature list FC_h ＝{FC_h0 ,……，FC_hk ，……，FC_hμ }，FC_hk ＝{FC¹_hk ，……，FC^θ_hk ，……，FC^ζ_hk }，FC^θ_hk Is C_hk Corresponding to the first record included in the θ -th first field in the first image feature list, θ= … … ζ, where ζ is the number of first fields.

4. A data processing system for acquiring facial image features as in claim 3, comprising the steps of, in S505:

s5051, when F^iu_hβ ≥F₀ At the time, F is acquired^iu_hβ Corresponding C'^u_hβ ；

S5053, when C'^u_hβ ＝C_hk At the time, the AC is acquired_i ＝FC_hk 。

5. A data processing system for acquiring facial image features according to claim 3, wherein the first image data processing platform further comprises a second image feature list M ' = { M ' corresponding to M '₁ ，……，M'_s ，……，M'_v }，M'_s ＝{M'_s1 ，……，M'_sε ，……，M'_sη }，M'_sε Is M_s And the epsilon=1 … … eta, eta is the number of the second fields, of the second records contained in the epsilon second field in the corresponding second image feature list.

6. The data processing system for acquiring facial image features according to claim 5 wherein the fields contained in the first field belong to the fields in the second field.

7. The data processing system for acquiring facial image features according to claim 5, comprising the steps of, in S600:

s601, when the theta first field is a preset field and the epsilon second field is a preset field, obtaining the FC^θ_hk And M'_sε Wherein the preset field is an ID;

s603, when s=h+k×n, acquiring AD_i ＝M'_s 。