Method, device and electronic device for searching pictures

By performing feature extraction and similarity calculation of search images, and selecting archive data sets as target images, the problem of wasted computing resources in massive portrait capture images is solved, and efficient target image search is achieved.

CN115017359BActive Publication Date: 2024-12-03ZHEJIANG DAHUA TECH CO LTD
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202210591291.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-05-27
Publication Date
2024-12-03
Estimated Expiration
2042-05-27

AI Technical Summary

Technical Problem

Among the massive portrait capture images, the prior art requires traversal and calculating the similarity between the image to be searched and each portrait capture image, resulting in serious waste of calculation time and resources.

Method used

Feature extraction is performed by treating the search image, calculate the reference individual feature similarity values ​​of each of the N archive data sets, and select the archive data set corresponding to the similarity value greater than the preset threshold as the target image to reduce the search data volume and improve the search speed .

Benefits of technology

There is no need to traverse each database picture, searching based on reference individual features only, saving computing resources, and improving search speed and accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115017359B_ABST
    Figure CN115017359B_ABST
Patent Text Reader

Abstract

The present application provides a method, device and electronic device for searching pictures. The method includes extracting features from a picture to be searched containing a target object, obtaining individual features of the target object, and then respectively calculating similarity values ​​between the individual features of the target object and I reference individual features corresponding to each of N archive data sets, obtaining N×I similarity values, where N and I are integers greater than 0, a single archive data set includes at least one or more pictures of the same object, and the reference individual features are obtained based on the individual features of the object contained in the picture in the archive data set, and finally selecting the archive data set corresponding to the similarity value greater than a preset threshold among the N×I similarity values, and using the picture in the selected archive data set as the target picture of the picture to be searched. Based on the above method, the amount of search data in the search process can be reduced, the computing resources used for the search can be saved, and the response speed of the search can be improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of image search, and in particular to a method, device and electronic device for searching images. Background Art

[0002] With the popularization of video surveillance equipment, portrait snapshots taken by video surveillance equipment are generated every moment. Based on this, among these portrait snapshots, a portrait snapshot image matching the specified target object can be searched according to a to-be-searched image of a specified target object as a target image corresponding to the to-be-searched image.

[0003] Currently, the search image is generally compared with each portrait snapshot image one by one for similarity, and the portrait snapshot image whose similarity meets certain conditions is selected as the target image.

[0004] However, when faced with a massive amount of portrait snapshot images, based on the above method, each time a target image is searched, it is necessary to traverse and calculate the similarity between the image to be searched and all the portrait snapshot images. This traversal calculation will not only waste a lot of computing time, but also consume a lot of computing resources. Summary of the invention

[0005] The present application provides a method, device and electronic device for searching an image, which are used to search for a target image that matches a target object in an image to be searched.

[0006] In a first aspect, the present application provides a method for searching for images, the method comprising:

[0007] Extracting features from the image to be searched that contains the target object to obtain individual features of the target object;

[0008] Calculate the similarity values ​​between the individual features of the target object and the I reference individual features corresponding to each of the N archive data sets, respectively, to obtain N×I similarity values; wherein N and I are integers greater than 0, a single archive data set includes at least one or more pictures of the same object, and the reference individual features are obtained based on the individual features of the object contained in the pictures in the archive data set;

[0009] An archive data set corresponding to a similarity value greater than a preset threshold value among the N×I similarity values ​​is selected, and a picture in the selected archive data set is used as a target picture of the picture to be searched.

[0010] Through the above method, when searching for a target image containing a target object, it is not necessary to traverse and search each image in the database, but only based on the I reference individual features corresponding to each of the N archive data sets, the target object can be searched, and the archive data set matching the target object can be determined, and the image in the archive data set can be used as the target image. Based on this, the amount of search data in the search process can be reduced, thereby saving the computing resources used in the search process and improving the search speed of the target image.

[0011] In one possible design, the method further includes: based on the individual characteristics of the objects contained in the pictures in the data set, dividing the pictures containing the same object into the same archival data set to obtain N archival data sets divided by the data set; and calculating the reference individual characteristics corresponding to the single archival data set according to the individual characteristics corresponding to each picture in the single archival data set to obtain the reference individual characteristics corresponding to each of the N archival data sets.

[0012] Through the above method, after the images of the same object are divided into the same archive data set, the reference individual features corresponding to the same archive data set are determined according to the individual features of the objects contained in the images in the same data set. The reference individual features calculated in this way can better characterize the same object contained in the images in the same archive data set. Therefore, the target image determined based on the reference individual features can also be more accurate. On the other hand, it is not necessary to determine the target image based on the individual features of each image in the archive data set. It is only necessary to calculate the corresponding reference individual features for each archive data set, which can effectively reduce the computing resources consumed and the wasted computing time for the subsequent search for the target image.

[0013] In a possible design, the dividing of pictures containing the same object into the same archival data set based on individual features of the objects contained in the pictures in the data set includes: calculating an individual feature similarity value, a time stamp similarity value, and a space stamp similarity value between two pictures in the data set; wherein the time stamp similarity value represents the interval between the shooting times of the two pictures, and the space stamp similarity value represents the distance between the shooting locations of the two pictures; in response to the individual feature similarity value being greater than a first threshold, dividing the two pictures into the same archival data set; in response to the individual feature similarity value being less than or equal to the first threshold and greater than a second threshold, the time stamp similarity value being greater than a third threshold, and the space stamp similarity value being greater than a fourth threshold, dividing the two pictures into the same archival data set; wherein the second threshold is less than the first threshold.

[0014] Through the above method, a method for dividing images to obtain archive data sets by combining the spatial dimension and time dimension of the image is proposed, which solves the problem that in some scenarios, when object A passes through adjacent cameras A and camera B within a certain period of time, and the individual feature similarity value of the images of object A collected by cameras A and camera B is less than the first threshold, the images of object A collected by cameras A and camera B are considered to be images of different objects. To solve this problem, for the images whose individual feature similarity value meets the second threshold, the interval of the collection time and the distance of the collection position that meet certain conditions are also regarded as images corresponding to the same object, so as to improve the accuracy of generating archive data sets and the filing rate of images included in the archive data sets. In addition, improving the filing rate of images included in the archive data sets is also conducive to reducing the computing resources and computing time required for the subsequent search for target images.

[0015] In a possible design, the pictures containing the same object are divided into the same archival data set based on individual characteristics of the objects contained in the pictures in the data set to obtain N archival data sets divided by the data set, including: among the pictures in the data set, pictures with picture quality values ​​greater than a preset quality threshold are divided as a first data set, and pictures with picture quality values ​​less than or equal to the preset quality threshold are divided as a second data set; for the first data set and the second data set, pictures containing the same object are divided into the same archival data set respectively to obtain multiple first archival data sets divided by the first data set and multiple second archival data sets divided by the second data set; the multiple first archival data sets and the multiple second archival data sets are collectively used as the N archival data sets divided by the data set.

[0016] Through the above method, a method for generating an archive data set based on the division of the picture quality dimension is proposed. Based on the picture quality value of the picture, the pictures with higher picture quality values ​​are taken as the first data set, and the pictures with lower picture quality values ​​are taken as the second data set. Then, the pictures in the first data set and the second data set are clustered respectively. It is obvious that the clustering results obtained by clustering in this way are more accurate than the clustering results directly targeting the data set without division. Based on this, the accuracy of the N archive data sets generated by clustering can be improved, thereby improving the accuracy of the target pictures searched based on these N archive data sets.

[0017] In one possible design, the calculating of reference individual features corresponding to the single archival data set based on the individual features corresponding to each picture in the single archival data set includes: calculating the average features of the individual features corresponding to each picture in the single archival data set, and using the average features as the reference individual features corresponding to the single archival data set; and / or determining the individual features corresponding to a specified picture in the single archival data set, and using the determined individual features as the reference individual features corresponding to the single archival data set.

[0018] Through the above method, a method for calculating reference individual characteristics is proposed. Regardless of whether the average characteristics of the archive data set and / or the individual characteristics corresponding to the specified image are used as the reference individual characteristics of the archive data set, the reference individual characteristics calculated in this way can more significantly reflect the individual characteristics of the same object corresponding to the archive data set, thereby helping to improve the accuracy of the target image searched based on the reference individual characteristics.

[0019] In one possible design, the selecting of the archive data set corresponding to the similarity value greater than a preset threshold among the N×I similarity values ​​includes: among the N×I similarity values, using the similarity value corresponding to the first data set as the first similarity value, and using the similarity value corresponding to the second data set as the second similarity value; in response to the first similarity value being greater than a first preset threshold, selecting the archive data set corresponding to the first similarity value; in response to the first similarity values ​​being less than or equal to the first preset threshold, and the second similarity value being greater than a second preset threshold, selecting the archive data set corresponding to the second similarity value.

[0020] Through the above method, the first similarity value is calculated based on the reference individual features corresponding to the first archive data set with good image quality. If the first similarity value does not meet the first preset threshold, the first archive data is not the archive data containing the target object. Further, the second similarity value is calculated based on the reference individual features corresponding to the second archive data set with poor image quality. Based on this, firstly, the accuracy of searching for the target image can be improved, and secondly, the computing resources and computing time wasted in searching for the target image can be effectively saved.

[0021] In a possible design, after obtaining the reference individual features corresponding to each of the N archival data sets, it also includes: respectively calculating the similarity values ​​between the reference individual features corresponding to each of the N archival data sets and the reference individual features corresponding to the historical archival data sets; in response to the similarity value being greater than a preset similarity threshold, merging the archival data set corresponding to the similarity value with the historical archival data set to obtain a merged archival data set.

[0022] Through the above method, a situation where there is a historical archive data set is proposed. In this case, the historical archive data set containing the same object and the currently obtained archive data set are merged. This not only increases the number and accuracy of the searched target images, but also reduces the number of final participating archive data sets, thereby saving computing resources and computing time wasted in searching for target images.

[0023] In a second aspect, the present application provides a device for searching for images, the device comprising:

[0024] A feature extraction module extracts features from the image to be searched containing the target object to obtain individual features of the target object;

[0025] A similarity value calculation module calculates similarity values ​​between the individual features of the target object and I reference individual features corresponding to each of the N archive data sets, respectively, to obtain N×I similarity values; wherein N and I are integers greater than 0, a single archive data set includes at least one or more pictures of the same object, and the reference individual features are obtained based on the individual features of the object contained in the pictures in the archive data set;

[0026] The target image selection module selects an archive data set corresponding to a similarity value greater than a preset threshold among the N×I similarity values, and uses the image in the selected archive data set as the target image of the image to be searched.

[0027] In one possible design, the device further includes: a module for dividing an archival data set, which divides pictures containing the same object into the same archival data set based on the individual characteristics of the objects contained in the pictures in the data set, so as to obtain N archival data sets divided by the data set; and a module for calculating reference individual characteristics, which calculates the reference individual characteristics corresponding to the single archival data set according to the individual characteristics corresponding to each picture in the single archival data set, so as to obtain the reference individual characteristics corresponding to each of the N archival data sets.

[0028] In a possible design, the module for dividing the archive data set is specifically used to calculate the individual feature similarity value, time stamp similarity value and space stamp similarity value between two pictures in the data set; wherein the time stamp similarity value represents the interval between the shooting times of the two pictures, and the space stamp similarity value represents the distance between the shooting locations of the two pictures; in response to the individual feature similarity value being greater than a first threshold, the two pictures are divided into the same archive data set; in response to the individual feature similarity value being less than or equal to the first threshold and greater than a second threshold, the time stamp similarity value being greater than a third threshold, and the space stamp similarity value being greater than a fourth threshold, the two pictures are divided into the same archive data set; wherein the second threshold is less than the first threshold.

[0029] In a possible design, the module for dividing the archive data set is specifically used to divide the pictures in the data set, the pictures whose quality values ​​are greater than a preset quality threshold as the first data set, and the pictures whose quality values ​​are less than or equal to the preset quality threshold as the second data set; for the first data set and the second data set, respectively, the pictures containing the same object are divided into the same archive data set to obtain multiple first archive data sets divided by the first data set, and multiple second archive data sets divided by the second data set; the multiple first archive data sets and the multiple second archive data sets are collectively used as the N archive data sets divided by the data set.

[0030] In one possible design, the module for calculating reference individual features is specifically used to calculate the average features of individual features corresponding to each image in a single archival data set, and use the average features as the reference individual features corresponding to the single archival data set; and / or determine the individual features corresponding to a specified image in the single archival data set, and use the determined individual features as the reference individual features corresponding to the single archival data set.

[0031] In one possible design, the module for dividing the archive data set is specifically used to, among the N×I similarity values, use the similarity value corresponding to the first data set as the first similarity value, and use the similarity value corresponding to the second data set as the second similarity value; in response to the first similarity value being greater than a first preset threshold, select the archive data set corresponding to the first similarity value; in response to the first similarity values ​​being less than or equal to the first preset threshold, and the second similarity value being greater than the second preset threshold, select the archive data set corresponding to the second similarity value.

[0032] In one possible design, the device is further used to respectively calculate the similarity values ​​between the reference individual features corresponding to each of the N×I archival data sets and the reference individual features corresponding to the historical archival data set; in response to the similarity value being greater than a preset similarity threshold, the archival data set corresponding to the similarity value is merged with the historical archival data set to obtain a merged archival data set.

[0033] In a third aspect, the present application provides an electronic device, the electronic device comprising:

[0034] Memory, used to store computer programs;

[0035] The processor is used to implement the above-mentioned method steps for searching images when executing the computer program stored in the memory.

[0036] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned method steps for searching for images are implemented.

[0037] For each aspect from the second to the fourth aspect and the technical effects that may be achieved by each aspect, please refer to the above description of the technical effects that can be achieved by the first aspect or various possible schemes in the first aspect, and no further details will be given here. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 A schematic diagram of a possible application scenario provided for this application;

[0039] Figure 2 A flowchart of a method for determining reference individual characteristics provided in the present application;

[0040] Figure 3 A flowchart of a method for searching images provided in this application;

[0041] Figure 4 A schematic diagram of a device for searching images provided by the present application;

[0042] Figure 5 A schematic diagram of the structure of an electronic device provided in this application. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solution and advantages of the present application more clear, the present application will be further described in detail below in conjunction with the accompanying drawings. The specific operation method in the method embodiment can also be applied to the device embodiment.

[0044] In the description of this application, "multiple" is understood to be "at least two". "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent three situations: A exists alone, A and B exist at the same time, and B exists alone. A is connected to B, which can represent two situations: A is directly connected to B and A is connected to B through C. In addition, in the description of this application, words such as "first" and "second" are only used for the purpose of distinguishing descriptions, and cannot be understood as indicating or implying relative importance, nor can they be understood as indicating or implying order.

[0045] It is worth noting that in the description of this application, the collection / acquisition of the privacy information of the object shall comply with the relevant laws and regulations. In general, the collection / acquisition of the privacy information of the object requires the permission of the corresponding object. Among them, the object can generally be a person, an animal, or an object with common characteristics, and the privacy information can be a picture of the object, the individual information of the object parsed / acquired based on the picture, etc.

[0046] In addition, in the process of executing the method of this application, the use of the target object's private information should comply with relevant laws and regulations, including but not limited to, application in scenarios that help promote the public interest and positively promote social morality.

[0047] See also Figure 1 As shown, a possible application scenario provided by an embodiment of the present application specifically includes a server and a client.

[0048] exist Figure 1 In the process, the client is used to obtain a to-be-searched image containing a target object, and after the client obtains the to-be-searched image, the client is also used to obtain a search instruction and send the search instruction to the server.

[0049] The server stores the archive data set and the reference individual features corresponding to each archive data set. For example, the server may store N archive data sets and I reference individual features corresponding to each archive data set. After the server receives the search instruction sent by the client, the server compares the individual features of the target object with the I reference individual features corresponding to each archive data set based on the target object in the image to be searched, searches for the archive data set that matches the target object, uses the image in the archive data set as the target image, and then sends the searched archive data set and / or target image to the client.

[0050] The client is also used to receive the archive data set and / or target image sent by the server, and display the archive data set and / or target image in the visualization interface.

[0051] Based on the above application scenarios, the embodiments of the present application provide a method, device and electronic device for searching images, which solve the problems of the prior art in that searching images is time-consuming and wastes computing resources.

[0052] In addition, the technical features included in the embodiments of the present application can be used in any combination. Those skilled in the art should understand that, based on actual application conditions, the technical solutions obtained by reasonably combining the technical features in the embodiments of the present application can also solve the same technical problems or achieve the same technical effects.

[0053] According to the method provided in the embodiment of the present application, the individual features of the target object in the image to be searched are first determined, and then N×I similarity values ​​between the individual features and I reference individual features corresponding to each of N archival data sets are calculated, and finally, the image in the archival data set corresponding to the maximum similarity value is selected as the target image to be searched.

[0054] Specifically, each of the above-mentioned archive data sets includes at least one picture of the same object, and each archive data set corresponds to at least one reference individual feature. That is to say, by the method provided by the embodiment of the present application, it is not necessary to search each picture in the database ergodicly when searching for the target picture containing the target object, and it is only necessary to search for the target object based on the 1 reference individual features corresponding to each of these N archive data sets, and then determine the archive data set that matches the target object, and use the picture in the archive data set as the target picture. Therefore, by the method provided by the embodiment of the present application, the search data amount in the search process can be reduced, and then the computing resources used by the search process are saved, and the search speed of the search target picture is improved.

[0055] Furthermore, the search for the target image is completed by comparing it with the reference individual features, that is, the accuracy of the reference individual features will affect the accuracy of the target image finally searched out. Therefore, in order to improve the accuracy of the search for the target image, the following steps can also be performed before executing the method for searching images provided in the embodiments of the present application.

[0056] See also Figure 2 As shown, the present application provides a method for determining reference individual characteristics, and the specific process is as follows:

[0057] Step 201: based on individual features of objects contained in the images in the data set, images containing the same object are divided into the same archive data set to obtain N archive data sets divided from the data set;

[0058] Step 202: Calculate reference individual features corresponding to a single archival data set according to individual features corresponding to each image in the single archival data set, and obtain reference individual features corresponding to each of the N archival data sets.

[0059] In the embodiment of the present application, each picture in the data set contains a corresponding object, which may refer to a person, an animal, or an object, etc. The data set may be obtained from the storage data in the server, or from the image data collected by the camera, such as the image data taken by the fisheye camera, infrared camera, etc. It is understandable that the camera may also collect the image data in real time, such as the camera of the roadside monitoring equipment, etc.

[0060] It is worth noting that in order to facilitate technical personnel in this field to better understand the technical solutions provided in the embodiments of the present application, in the embodiments of the present application, people are taken as objects in the pictures, and portrait pictures taken by the camera within a specified time period are taken as pictures in the database, which will not be repeated below.

[0061] For the images in the data set, a feature extraction method can also be used to extract individual features of the portraits in the images. In the embodiment of the present application, the individual features of the portraits can include facial features and / or body features. For a portrait image 1 containing object A, if there is only the facial part of A in the portrait image 1, only the facial features of A can be extracted; if there is only the body part of A in the portrait image 1, only the body features of A can be extracted; if there is both the facial part and the body part of A in the portrait image 1, both the facial features and the body features of A can be extracted.

[0062] After determining that the pictures in the data set contain facial features and / or body features of the objects, N archival data sets can be obtained by performing clustering operations on the facial features and / or body features, that is, pictures whose facial features and / or body features meet certain similarity conditions are divided into the same archival data set.

[0063] Furthermore, in order to improve the clustering accuracy of images in the same archive data set in the embodiment of the present application, before performing a specific clustering operation, the images in the data set can also be divided into a first data set and a second data set based on the image quality values ​​of the images.

[0064] The image quality value may be related to the resolution and clarity of the image. For example, a larger image quality value may indicate a higher resolution and clarity of the image; conversely, a smaller image quality value may indicate a lower resolution and clarity of the image. Optionally, the image quality value may also be related to the size of the image.

[0065] Specifically, based on the image quality value of each image in the data set, images with image quality values ​​greater than a preset quality threshold are divided into the first data set; images with image quality values ​​less than or equal to the preset quality threshold are divided into the second data set. The objects contained in the images in the first data set obtained by this division method are clearer than the objects contained in the images in the second data set. Therefore, compared with the second data set, the facial features and / or body features corresponding to the images in the first data set are also more accurate.

[0066] Based on this, clustering operations are performed on the pictures in the first data set and the second data set respectively, to obtain a plurality of first archive data sets divided by the first data set, and a plurality of second archive data sets divided by the second data set. Because the facial features and / or body features corresponding to the pictures in the first data set are more accurate, and because the clustering operation needs to be performed based on these facial features and / or body features, clustering the first data set and the second data set separately can effectively improve the accuracy of clustering, thereby improving the clustering accuracy of the N archive data sets composed of the plurality of first archive data sets and the plurality of second archive data sets.

[0067] Optionally, in some possible implementations, the clustering operation may be performed only on the images in the first data set, and then a plurality of first archive data sets obtained by clustering the images in the first data set are used as the N archive data sets.

[0068] Further, in the embodiment of the present application, in order to improve the accuracy of the N archive data sets generated by clustering, the time dimension and the space dimension can also be combined to perform clustering operations on the images. The following takes the clustering operation on the images in the first data set within the selected time period as an example, and makes the following detailed description.

[0069] It is worth noting that the images in the first data set can correspond to three situations: face features, body features, and face features and body features. Each situation can generate a corresponding feature file, such as a face file can be generated based on face features, a body file can be generated based on body features, and a face and body file can be generated based on face features and body features. Here, the face files, body files, and face and body files generated in these three situations can all be used as the first file data set. The following is a detailed explanation of the generation of the above face files, body files, and face and body files, divided into three situations.

[0070] Case 1: Generate a face file based on the picture corresponding to the facial features;

[0071] First, the time mark and space mark of the above picture can be determined. Here, the time mark is the shooting time (collection time) of the picture, and the space mark is the spatial position of the collection device of the picture. The spatial position can be the card mark, geographical longitude and latitude of the collection device, etc. The collection device is a device that at least includes a camera. Then, the individual feature similarity value between the facial features of the above two pictures, the time mark similarity value between the two pictures, and the space mark similarity between the two pictures are calculated.

[0072] Then, the calculated individual feature similarity value is compared with the first threshold: if the calculated individual feature similarity value is greater than the first threshold, it is considered that the two pictures contain the same object and the two pictures are divided into the same cluster. Here, a single cluster is an important part of a single face file; if the calculated individual feature similarity value is less than or equal to the first threshold, the size relationship between the individual feature similarity value and the second threshold is further determined. Here, the second threshold is less than the first threshold.

[0073] Further, it is determined whether the individual feature similarity value is greater than the second threshold: if the individual feature similarity value is less than or equal to the second threshold, it is determined that the two images contain different objects and the two images are divided into different clusters; if the individual feature similarity value is greater than the second threshold, it is also necessary to determine whether the time mark similarity value is greater than the third threshold, and to determine whether the spatial similarity value is greater than the fourth threshold. Here, if the time similarity value of the two images is greater than the third threshold, and the spatial similarity value is greater than the fourth threshold, it is determined that the two images contain the same object, and the two images are divided into the same cluster; otherwise, it is determined that the two images contain different objects, and the two images are divided into different clusters.

[0074] Through the above clustering operation, not only the similarity of the facial features of the two images is considered, but also the similarity of the time and space dimensions of the two images are combined. Therefore, the images in each cluster obtained based on the above method have time and space correlation.

[0075] For example, there is a camera at each intersection, and the walking time between two intersections is 1 minute. Assume that object 1 is walking on the road, and camera A at intersection A captures a picture A, and camera B at the next intersection B also captures a picture B. The interval between the acquisition time of pictures A and B is 1 minute, so even if the facial features of the two pictures are not very similar, pictures A and B can still be considered to be the same person. Therefore, combining the time dimension and the space dimension helps to improve the accuracy of image clustering, and thus helps to improve the accuracy of subsequent recognition of target pictures.

[0076] Furthermore, after obtaining multiple clusters by clustering the pictures in the first data set, a recommended centroid and a mean centroid can be calculated for each cluster. The recommended centroid can be the facial features of a specified picture or a representative picture in the cluster, and the mean centroid can be the average facial features of the facial features of all pictures in the cluster.

[0077] Optionally, this clustering operation obtains multiple clusters and the recommended centroid and / or mean centroid of each cluster. If there are still clusters from the previous clustering operation, the recommended centroids of the clusters obtained from this clustering operation can be compared one by one with the recommended centroids of the clusters obtained from the previous clustering operation to obtain the recommended centroid similarity value; and / or the mean centroids of the clusters obtained from this clustering operation can be compared one by one with the mean centroids of the clusters obtained from the previous clustering operation to obtain the mean centroid similarity value. The calculated recommended centroid similarity value and / or mean centroid similarity value are then compared with a preset threshold to determine whether the objects in the two clusters are the same object, suspected to be the same object, or different objects.

[0078] Taking the mean centroid similarity value as an example, it can be specifically set as follows: if the mean centroid similarity is less than 60, the objects in the two clusters are determined to be different objects; if the mean centroid similarity is greater than or equal to 60 and less than or equal to 90, the objects in the two clusters are determined to be suspected to be the same object; if the mean centroid similarity is greater than 90, the objects in the two clusters are determined to be the same object.

[0079] Furthermore, in order to improve the accuracy of clustering, clusters suspected of being the same object are compared one by one. Specifically, the facial features of all clusters suspected of being the same object are compared one by one to calculate the similarity values ​​between all the clusters. If the calculated similarity value meets the set threshold, it is determined that the images in these clusters contain the same object.

[0080] Based on the above method, further comparing features one by one in screening clusters of suspected identical objects can not only reduce the computation time and resources for comparing features, but also effectively improve the accuracy of clustering.

[0081] Furthermore, if it is determined that the objects in the two clusters are the same object, the two clusters can be merged to obtain a merged new cluster, and then the recommended centroid and / or mean centroid of the new cluster can be recalculated.

[0082] The above method is not only conducive to improving the accuracy of clustering, thereby improving the accuracy of subsequent image searches, but also can effectively reduce the number of clusters, thereby reducing the computing resources and computing time consumed by subsequent image searches.

[0083] Based on the above clustering method, the facial features of the images in the first data set are clustered to obtain multiple clusters and the mean centroid and recommended centroid corresponding to each cluster. Because each cluster represents the same object, in the embodiment of the present application, a corresponding face file can also be generated for each cluster, and the images in the cluster, the mean centroid and / or recommended centroid of the cluster are saved in the face file as reference individual features.

[0084] Optionally, the face file can also be divided into a face main file and a face sub-file: the face main file can be used to represent the individual information corresponding to the face of the same object, which may specifically include the object's name, ID number, age group, gender, reference individual characteristics, etc.; the face sub-file can be used to represent the face image data of the same object in the picture, and the face image data may be based on the face information analyzed from the face image and the device information of the face image acquisition device. The face information may include the object's age, gender, facial features, etc., and the device information may include the acquisition time of the acquisition device, the location of the acquisition device, the device ID of the acquisition device, etc.

[0085] Furthermore, if the face main file and face sub-file of the object are generated based on the cluster, then the association between the face main file and face sub-file of the same object will be established to form associated data stored in the corresponding face main file and face sub-file.

[0086] Case 2: Generate a human profile based on the pictures corresponding to human features;

[0087] In the embodiment of the present application, the images in the first data set are clustered based on human body features. The clustering method for image clustering is described in the first case and will not be described in detail here.

[0088] Case 2 is similar to Case 1. Based on the clustering operation, multiple human profiles and reference individual features of each human profile can be generated. However, it is worth noting that the reference individual features in Case 2 are obtained based on human features.

[0089] Optionally, the human body file can also be divided into a human body main file and a human body sub-file. Among them, the human body main file is used to represent the individual information corresponding to the human body of the same object, which can specifically include the name, ID number, age group, gender, reference individual characteristics, etc. of the object; the human body sub-file can be used to represent the human body image data of the same object in the picture, and the face image data can be based on the human body information obtained by parsing the human body image and the device information of the human body image acquisition device. The human body information can include the front, side and back information of the object, glasses information, hair color information, etc., and the device information can include the acquisition time of the acquisition device, the location of the acquisition device, the device ID of the acquisition device, etc.

[0090] Case 3: Generate face and body files based on pictures with corresponding facial features and body features.

[0091] In an embodiment of the present application, a clustering operation based on case one is first obtained to obtain multiple clusters, and the pictures containing facial features and body features are divided into clusters according to the obtained cluster grouping method to obtain multiple newly divided clusters. Then, the recommended centroid and / or mean centroid of each newly divided cluster is calculated. The calculation method refers to the relevant description of case one and is not elaborated in detail here. A corresponding human body profile is generated for each newly divided cluster, and the recommended centroid and / or mean centroid corresponding to the cluster is used as the reference individual feature corresponding to the human body profile.

[0092] Furthermore, since the human body profile in case 3 is generated based on the face profile obtained in case 1, an association relationship can also be established between the human body profile in case 3 and the face profile in case 1. Based on the established association relationship, a face and human body profile is generated. The face and human body profile includes a face profile and a human body profile of the same object and an association relationship between the face profile and the human body profile.

[0093] Through the above three situations, the face file, body file and face-body file of the pictures in the first data set can be generated. Here, each generated face file, body file or face-body file can be used as the first archive data set of the first data set. Then, the first archive data set consisting of the face file, body file and face-body file corresponding to the first data set, and the reference individual features corresponding to each first archive data set are obtained.

[0094] In the embodiment of the present application, based on the same processing method, a second archive data set of the second data set and reference individual features corresponding to each second archive data set can also be obtained, which will not be elaborated in detail here.

[0095] Through the above method, based on the individual characteristics of the objects contained in the pictures in the data set, the pictures containing the same object are divided into the same archival data set to obtain N archival data sets divided by the data set. Then, according to the individual characteristics corresponding to each picture in the single archival data set, the reference individual characteristics corresponding to the single archival data set are calculated to obtain the reference individual characteristics corresponding to each of the N archival data sets.

[0096] It is worth noting that the above-mentioned method for determining reference individual characteristics is a preferred implementation mode provided by the present application. Those skilled in the art should understand that, based on actual application situations, the technical solution obtained by reasonably splitting and combining the technical features in steps 101 and 102 of the present application can also solve the same technical problems or achieve the same technical effects.

[0097] Based on the above method for determining the reference individual features corresponding to each of the N archival data sets, the method provided in the embodiment of the present application is further described in detail below in conjunction with the accompanying drawings.

[0098] See also Figure 3 As shown, the embodiment of the present application provides a method for searching for pictures, and the specific process is as follows:

[0099] Step 301: extracting features from a to-be-searched image containing a target object to obtain individual features of the target object;

[0100] In the embodiment of the present application, a picture to be searched containing a target object is first obtained, and then individual features of the target object are extracted from the picture to be searched to extract the individual features of the target object.

[0101] The target object can be used to characterize a specific individual, that is, the target object can be a person, a creature, an object, etc. The individual characteristics of the target object can be used to identify the target object, that is, when the target object is person A, A can be identified from multiple different people including A based on the individual characteristics of A.

[0102] Furthermore, when the object is a person, the individual characteristics of the person may specifically be facial characteristics, body characteristics, and other biological characteristics of the person. It is worth noting that in the embodiment of the present application, the following specific description is made by taking the case that the image to be searched contains a person and the individual characteristics of the person include facial characteristics and / or body characteristics, and no further description will be given below.

[0103] Taking the example that individual features can include facial features and body features, the feature extraction for the image to be searched will be divided into the following three cases: when the image to be searched only contains the face of the target object, the feature extraction will only obtain the facial features of the target object; when the image to be searched only contains the body of the target object, the feature extraction will only obtain the body features of the target object; when the image to be searched contains both the face and body of the target object, the feature extraction will obtain the facial features and body features of the target object. Here, the extracted facial features and body features are both individual features of the target object.

[0104] Step 302: Calculate the similarity values ​​between the individual features of the target object and the I reference individual features corresponding to each of the N archive data sets to obtain N×I similarity values;

[0105] After extracting the individual features corresponding to the image to be searched, it is also necessary to obtain the I reference individual features corresponding to each of the N archive data sets, that is, to obtain N×I reference individual features, and then respectively calculate the similarity values ​​between the individual features and the N reference individual features, to obtain N×I similarity values. Here, the similarity value can be used to indicate the degree of similarity between the target object corresponding to the image to be searched and the object corresponding to the archive data set.

[0106] In an embodiment of the present application, N and I are integers greater than 0, a single archival data set includes at least one or more pictures of the same object, and a single archival data set corresponds to at least one reference individual feature, which can be obtained based on the individual features of the object contained in the picture in the archival data set.

[0107] Specifically, because the same archive data set contains pictures of the same object, the reference individual features corresponding to the same archive data set can be obtained based on the individual features of the same object. For example, in archive data set A, there are pictures 1 and 2 containing object a. Picture 1 extracts individual feature 1 of object a, and picture 2 extracts individual feature 2 of object a. Here, the average value of individual feature 1 and individual feature 2 can be used as the reference individual feature corresponding to archive data set A, or the specified individual feature 1 or individual feature 2 can be used as the reference individual feature corresponding to archive data set A.

[0108] It is worth noting that the above-mentioned individual characteristics and the above-mentioned reference individual characteristics can be expressed as characteristic vectors or as characteristic values. However, no matter whether it is a characteristic vector or an characteristic value, it should be ensured that the data form of the individual characteristics and the reference individual characteristics when participating in mathematical calculations is the same. In some possible implementations, if the individual characteristics and the reference individual characteristics are both expressed as characteristic vectors, then the similarity value between the individual characteristics and the reference individual characteristics is calculated based on the similarity between the two characteristic vectors. In some possible implementations, if the individual characteristics and the reference individual characteristics are both expressed as characteristic values, the characteristic values ​​here can be the characteristic values ​​corresponding to the characteristic vectors, or can be the extracted characteristic values, then the similarity value between the individual characteristics and the reference individual characteristics is calculated based on the similarity between the two characteristic values.

[0109] Furthermore, if only facial features are extracted from the image to be searched, all facial archives can be selected as the N archive data sets, and then the similarity values ​​between the individual features of the target object and the I reference individual features corresponding to each of the N archive data sets are calculated to obtain N×I similarity values. In this way, the computing resources and computing time required for calculating the similarity values ​​can be further reduced.

[0110] If only facial features are extracted from the searched image, all facial files can be selected as the N file data sets, and then the similarity values ​​between the individual features of the target object and the I reference individual features corresponding to each of the N file data sets are calculated to obtain N×I similarity values. In this way, the computing resources and computing time required to calculate the similarity values ​​can be further reduced.

[0111] If the face features and body features are extracted from the search image, all face and body files can be selected as the N file data sets, and then the similarity values ​​between the individual features of the target object and the I reference individual features corresponding to each of the N file data sets are calculated to obtain N×I similarity values. In this way, the computing resources and computing time required for calculating the similarity values ​​can be further reduced.

[0112] Optionally, the above-mentioned face archive, body archive, face and body archive may be archives obtained by clustering based on the first data set.

[0113] Step 303: Select an archive data set corresponding to a similarity value greater than a preset threshold value among the N×I similarity values, and use the image in the selected archive data set as the target image of the image to be searched.

[0114] After calculating N×I similarity values ​​in step 302, the archive data set corresponding to the similarity value greater than the preset threshold is selected as the target image of the current image to be searched. Of course, the image in the archive data set corresponding to the maximum similarity value can also be directly used as the target image.

[0115] In addition, in an embodiment of the present application, these N×I similarity values ​​can be divided into I groups according to the categories of I reference individual features, that is, each group contains N similarity values, and then the pictures in the archive data set corresponding to the similarity value / maximum similarity value greater than a preset threshold in the specified group are selected as the target pictures.

[0116] In some possible implementations, the N archive data sets are first archive data sets corresponding to the first data set, and the objects contained in the pictures in the first archive data set do not match the target objects. In this case, the second archive data set corresponding to the second data can be used as the N archive data sets, and the operations of step 202-step 203 are performed again based on the updated N archive data sets to obtain the target picture of the picture to be searched. Of course, if the target picture is not searched out this time, it can be considered that the target picture does not exist, and a response message indicating that the target picture does not exist is generated.

[0117] Furthermore, after the target image is searched out, the image to be searched can also be added to the corresponding archive data set.

[0118] Through the method of the embodiment of the present application, the computing resources consumed in the process of searching for target images can be reduced, and the response time of searching for target images can be effectively improved.

[0119] Based on the same inventive concept, the present application also provides a device for searching images, which is used to search for target images that match the target objects in the images to be searched, and solves the problem that the existing search images use traversal calculations to waste computing time and computing resources. The target image is determined based on the similarity between the individual features of the target object in the image to be searched and the I reference individual features corresponding to each of the N archive data sets. Based on this, the amount of search data in the search process is reduced, and the computing resources used in the search process are saved, and the technical effect of improving the search speed of the target image is achieved. See Figure 4 , the device comprises:

[0120] The feature extraction module 401 extracts features from the image to be searched containing the target object to obtain individual features of the target object;

[0121] A similarity calculation module 402 calculates similarity values ​​between the individual features of the target object and I reference individual features corresponding to each of the N archive data sets, respectively, to obtain N×I similarity values; wherein N and I are integers greater than 0, a single archive data set includes at least one or more pictures of the same object, and the reference individual features are obtained based on the individual features of the object included in the pictures in the archive data set;

[0122] The target image selection module 403 selects an archive data set corresponding to a similarity value greater than a preset threshold value among the N×I similarity values, and uses the image in the selected archive data set as the target image of the image to be searched.

[0123] In one possible design, the device further includes: a module for dividing an archival data set, which divides pictures containing the same object into the same archival data set based on the individual characteristics of the objects contained in the pictures in the data set, so as to obtain N archival data sets divided by the data set; and a module for calculating reference individual characteristics, which calculates the reference individual characteristics corresponding to the single archival data set according to the individual characteristics corresponding to each picture in the single archival data set, so as to obtain the reference individual characteristics corresponding to each of the N archival data sets.

[0124] In a possible design, the module for dividing the archive data set is specifically used to calculate the individual feature similarity value, time stamp similarity value and space stamp similarity value between two pictures in the data set; wherein the time stamp similarity value represents the interval between the shooting times of the two pictures, and the space stamp similarity value represents the distance between the shooting locations of the two pictures; in response to the individual feature similarity value being greater than a first threshold, the two pictures are divided into the same archive data set; in response to the individual feature similarity value being less than or equal to the first threshold and greater than a second threshold, the time stamp similarity value being greater than a third threshold, and the space stamp similarity value being greater than a fourth threshold, the two pictures are divided into the same archive data set; wherein the second threshold is less than the first threshold.

[0125] In a possible design, the module for dividing the archive data set is specifically used to divide the pictures in the data set, the pictures whose quality values ​​are greater than a preset quality threshold as the first data set, and the pictures whose quality values ​​are less than or equal to the preset quality threshold as the second data set; for the first data set and the second data set, respectively, the pictures containing the same object are divided into the same archive data set to obtain multiple first archive data sets divided by the first data set, and multiple second archive data sets divided by the second data set; the multiple first archive data sets and the multiple second archive data sets are collectively used as the N archive data sets divided by the data set.

[0126] In one possible design, the module for calculating reference individual features is specifically used to calculate the average features of individual features corresponding to each image in a single archival data set, and use the average features as the reference individual features corresponding to the single archival data set; and / or determine the individual features corresponding to a specified image in the single archival data set, and use the determined individual features as the reference individual features corresponding to the single archival data set.

[0127] In one possible design, the module for dividing the archive data set is specifically used to, among the N×I similarity values, use the similarity value corresponding to the first data set as the first similarity value, and use the similarity value corresponding to the second data set as the second similarity value; in response to the first similarity value being greater than a first preset threshold, select the archive data set corresponding to the first similarity value; in response to the first similarity values ​​being less than or equal to the first preset threshold, and the second similarity value being greater than the second preset threshold, select the archive data set corresponding to the second similarity value.

[0128] In one possible design, the device is further used to respectively calculate the similarity values ​​between the reference individual features corresponding to each of the N×I archival data sets and the reference individual features corresponding to the historical archival data set; in response to the similarity value being greater than a preset similarity threshold, the archival data set corresponding to the similarity value is merged with the historical archival data set to obtain a merged archival data set.

[0129] Based on the above device, when searching for a target image containing a target object, it is not necessary to traverse and search each image in the database, but only based on the I reference individual features corresponding to each of the N archive data sets, the target object can be searched, and then the archive data set matching the target object is determined, and the image in the archive data set is used as the target image. Based on this, the amount of search data in the search process can be reduced, thereby saving the computing resources used in the search process and improving the search speed of the target image.

[0130] Based on the same inventive concept, an electronic device is also provided in an embodiment of the present application, and the electronic device can realize the function of the aforementioned device for searching pictures, referring to Figure 5 , the electronic device comprises:

[0131] At least one processor 501, and a memory 502 connected to the at least one processor 501. The specific connection medium between the processor 501 and the memory 502 is not limited in the embodiment of the present application. Figure 5 In the example, the processor 501 and the memory 502 are connected via a bus 500. The bus 500 is Figure 5The connections between other components are shown in bold lines, and are not intended to be limiting. The bus 500 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 Only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus. Alternatively, the processor 501 can also be called a controller, and there is no limitation on the name.

[0132] In the embodiment of the present application, the memory 502 stores instructions that can be executed by at least one processor 501. The at least one processor 501 can execute the image search method discussed above by executing the instructions stored in the memory 502. The processor 501 can implement Figure 4 The functions of each module in the device shown.

[0133] Among them, the processor 501 is the control center of the device, and can use various interfaces and lines to connect the various parts of the entire control device. By running or executing instructions stored in the memory 502 and calling data stored in the memory 502, the various functions of the device and processing data, the device can be monitored as a whole.

[0134] In one possible design, the processor 501 may include one or more processing units, and the processor 501 may integrate an application processor and a modem processor, wherein the application processor mainly processes an operating system, a user interface, and application programs, and the modem processor mainly processes wireless communications. It is understandable that the modem processor may not be integrated into the processor 501. In some embodiments, the processor 501 and the memory 502 may be implemented on the same chip, and in some embodiments, they may also be implemented separately on separate chips.

[0135] Processor 501 can be a general-purpose processor, such as a central processing unit (CPU), a digital signal processor, an application-specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, and can implement or execute the methods, steps and logic block diagrams disclosed in the embodiments of the present application. A general-purpose processor can be a microprocessor or any conventional processor, etc. The steps of the image search method disclosed in the embodiments of the present application can be directly embodied as a hardware processor, or can be executed by a combination of hardware and software modules in the processor.

[0136] The memory 502 is a non-volatile computer-readable storage medium that can be used to store non-volatile software programs, non-volatile computer executable programs and modules. The memory 502 may include at least one type of storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory, a random access memory (Random Access Memory, RAM), a static random access memory (Static Random Access Memory, SRAM), a programmable read-only memory (Programmable Read Only Memory, PROM), a read-only memory (Read Only Memory, ROM), an electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, EEPROM), a magnetic memory, a disk, an optical disk, etc. The memory 502 is any other medium that can be used to carry or store a desired program code in the form of an instruction or data structure and can be accessed by a computer, but is not limited thereto. The memory 502 in the embodiment of the present application can also be a circuit or any other device / system that can realize a storage function, for storing program instructions and / or data.

[0137] By designing and programming the processor 501, the code corresponding to the image search method described in the above embodiment can be fixed into the chip, so that the chip can execute the image search method when it is running. Figure 3 The steps of the picture searching method of the embodiment shown are as follows: How to design and program the processor 501 is a technique known to those skilled in the art and will not be described in detail here.

[0138] Based on the same inventive concept, an embodiment of the present application also provides a storage medium, which stores computer instructions. When the computer instructions are executed on a computer, the computer executes the image search method discussed above.

[0139] In some possible implementations, various aspects of the image search method provided in the present application may also be implemented in the form of a program product, which includes a program code. When the program product is run on an apparatus, the program code is used to enable the control device to execute the steps of the image search method according to various exemplary implementations of the present application described above in this specification.

[0140] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, devices / systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0141] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0142] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0143] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0144] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for searching images, characterized in that: The method comprises: Extracting features from the image to be searched that contains the target object to obtain individual features of the target object; Calculate the similarity values ​​between the individual features of the target object and the I reference individual features corresponding to each of the N archive data sets, respectively, to obtain N×I similarity values; wherein N and I are integers greater than 0, a single archive data set includes at least one or more pictures of the same object, and the reference individual features are obtained based on the individual features of the object contained in the pictures in the archive data set; Among the N×I similarity values, a similarity value corresponding to the first data set is used as a first similarity value, and a similarity value corresponding to the second data set is used as a second similarity value, wherein the picture quality value of the picture in the first data set is greater than a preset quality threshold, and the picture quality value of the picture in the second data set is less than or equal to the preset quality threshold; In response to the first similarity value being greater than a first preset threshold, selecting the archive data set corresponding to the first similarity value; In response to the first similarity values ​​being less than or equal to the first preset threshold, and the second similarity value being greater than a second preset threshold, selecting the archive data set corresponding to the second similarity value; Select an image from the archive data set as the target image to be searched.

2. The method according to claim 1, characterized in that The method further comprises: Based on individual features of objects contained in the images in the data set, images containing the same object are divided into the same archive data set to obtain N archive data sets divided by the data set; According to the individual features corresponding to each image in a single archival data set, the reference individual features corresponding to the single archival data set are calculated to obtain the reference individual features corresponding to each of the N archival data sets.

3. The method according to claim 2, characterized in that The step of dividing the images containing the same object into the same archive data set based on the individual features of the objects contained in the images in the data set includes: Calculate the individual feature similarity value, time mark similarity value and space mark similarity value between two pictures in the data set; wherein the time mark similarity value represents the interval between the shooting times of the two pictures, and the space mark similarity value represents the distance between the shooting locations of the two pictures; In response to the individual feature similarity value being greater than a first threshold, classifying the two images into the same archive data set; In response to the individual feature similarity value being less than or equal to the first threshold and greater than the second threshold, the time identifier similarity value being greater than the third threshold, and the space identifier similarity value being greater than the fourth threshold, the two pictures are divided into the same archive data set; wherein the second threshold is less than the first threshold.

4. The method according to claim 2, characterized in that The method of dividing the pictures containing the same object into the same archive data set based on the individual features of the objects contained in the pictures in the data set, and obtaining N archive data sets divided by the data set, includes: Dividing the pictures of the data set into the first data set and the second data set; For each of the first data set and the second data set, the pictures containing the same object are divided into the same archive data set to obtain a plurality of first archive data sets divided from the first data set and a plurality of second archive data sets divided from the second data set; The plurality of first archive data sets and the plurality of second archive data sets are collectively used as the N archive data sets divided into the data sets.

5. The method according to claim 2, characterized in that The step of calculating the reference individual features corresponding to the single archive data set according to the individual features corresponding to each image in the single archive data set includes: Calculate the average feature of the individual features corresponding to each image in a single archival data set, and use the average feature as the reference individual feature corresponding to the single archival data set; and / or Determine individual features corresponding to the specified image in the single archive data set, and use the determined individual features as reference individual features corresponding to the single archive data set.

6. The method according to any one of claims 2 to 5, characterized in that: After obtaining the reference individual features corresponding to each of the N archive data sets, the method further includes: Calculate the similarity values ​​between the reference individual features corresponding to each of the N archive data sets and the reference individual features corresponding to the historical archive data set; In response to the similarity value being greater than a preset similarity threshold, the archive data set corresponding to the similarity value is merged with the historical archive data set to obtain a merged archive data set.

7. A device for searching pictures, characterized in that: The device comprises: A feature extraction module extracts features from the image to be searched containing the target object to obtain individual features of the target object; A similarity value calculation module calculates similarity values ​​between the individual features of the target object and I reference individual features corresponding to each of the N archive data sets, respectively, to obtain N×I similarity values; wherein N and I are integers greater than 0, a single archive data set includes at least one or more pictures of the same object, and the reference individual features are obtained based on the individual features of the object contained in the pictures in the archive data set; Selecting a target image module, and taking a similarity value corresponding to the first data set as a first similarity value and a similarity value corresponding to the second data set as a second similarity value among the N×I similarity values, wherein the image quality value of the image in the first data set is greater than a preset quality threshold, and the image quality value of the image in the second data set is less than or equal to the preset quality threshold; In response to the first similarity value being greater than a first preset threshold, selecting the archive data set corresponding to the first similarity value; In response to the first similarity values ​​being less than or equal to the first preset threshold, and the second similarity value being greater than a second preset threshold, selecting the archive data set corresponding to the second similarity value; Select an image from the archive data set as the target image to be searched.

8. An electronic device, characterized in that: include: Memory, used to store computer programs; A processor, configured to implement the method steps of any one of claims 1 to 6 when executing the computer program stored in the memory.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method steps described in any one of claims 1 to 6 are implemented.

Citation Information

Patent Citations

  • Picture archiving method and device, terminal equipment and computer readable storage medium

    CN113918510A

  • Portrait file gathering method and device, electronic equipment and storage medium

    CN114357216A

  • Target person searching method and device, computer equipment and storage medium

    CN114491148A