WO2015101289A1

WO2015101289A1 - Image management method, apparatus and system

Info

Publication number: WO2015101289A1
Application number: PCT/CN2014/095584
Authority: WO
Inventors: 黄飞跃; 郑志昊; 梁柱; 黄琴; 吴永坚
Original assignee: 腾讯科技（深圳）有限公司
Priority date: 2013-12-30
Filing date: 2014-12-30
Publication date: 2015-07-09
Also published as: CN104750737A; CN104750737B

Abstract

Embodiments of the present invention provide an image management method, an apparatus and a system, wherein the method comprises: obtaining an image classification feature that is set; using the image classification feature, performing retrieval in images to be managed, and obtaining an image that is provided with the image classification feature, and classifying the image obtained into a category. Using a technical scheme provided by the embodiments of the present invention, a user does not need to manually choose an image from a large number of images, easing manual workload of the user and reducing operation complexity for the user.

Description

Picture management method, device and system

The present application claims priority to Chinese Patent Application No. 2013-10746306.7, entitled "A Photo Album Management Method and Apparatus", filed on Dec. 30, 2013, the entire contents of .

Technical field

The present invention relates to the field of communications technologies, and in particular, to a picture management method, apparatus, and system.

Background technique

In the online photo album, the number of pictures is increasing, so there is a need to manage the pictures in the web album. For example, you need to classify existing pictures. At present, the classification of pictures is realized by manual method. The specific process is as follows: If the user wants to classify and display the photos in his album according to different characters, he can only manually create the sub-album of different characters, and then Manually find the corresponding characters and move their photos to the corresponding sub-album. . The above technical solution is that the user performs manual classification and management operations completely, and the system does not assist the user to simplify the operation process. Therefore, the scheme is inefficient in managing pictures.

Summary of the invention

The embodiment of the invention provides a picture management method, device and system to solve the above technical problem.

A picture management method, including:

Obtaining the set image classification feature;

Using the classification feature to retrieve a picture to be managed to obtain a picture having the classification feature;

Classify the images obtained.

A picture management apparatus includes one or more processors and a storage medium storing operation instructions, and when operating an operation instruction in the storage medium, the processor performs the following steps:

Obtaining the set image classification feature;

Using the picture classification feature to search for a managed picture to obtain a picture having the picture classification feature;

Classify the images obtained.

A picture management system comprising:

Database server, annotation and recommendation server,

The database server is configured to store a picture to be processed;

And the labeling and recommendation server is configured to receive the set image classification feature, use the image classification feature to retrieve the image in the database server, obtain a picture having the image classification feature, and send the obtained image to the terminal. A picture for the terminal to select.

As can be seen from the above technical solutions, the embodiment of the present invention has the following advantages: after acquiring the image classification feature, the image to be managed is retrieved, and the image having the image classification feature is obtained. With the technical solution provided by the embodiment of the present invention, the user does not need to manually select a picture from a large number of pictures, which reduces the manual workload of the user and reduces the user operation complexity.

DRAWINGS

FIG. 1 is a schematic flowchart of a picture management method according to an embodiment of the present invention;

2 is a schematic flowchart of another picture management method according to an embodiment of the present invention;

3 is a schematic structural diagram of a user annotation and system recommendation system according to an embodiment of the present invention;

4 is a schematic structural diagram of a preprocessing system according to an embodiment of the present invention;

FIG. 5 is a schematic structural diagram of another labeling and recommendation system according to an embodiment of the present invention; FIG.

FIG. 6 is a schematic structural diagram of a picture management apparatus according to an embodiment of the present invention;

FIG. 7 is a schematic structural diagram of another picture management apparatus according to an embodiment of the present disclosure;

FIG. 8 is a schematic structural diagram of another picture management apparatus according to an embodiment of the present disclosure;

FIG. 9 is a schematic structural diagram of another picture management apparatus according to an embodiment of the present invention;

FIG. 10 is a schematic structural diagram of another picture management apparatus according to an embodiment of the present disclosure;

FIG. 11 is a schematic structural diagram of another picture management apparatus according to an embodiment of the present invention;

12 is a schematic structural diagram of a system according to an embodiment of the present invention;

13 is a schematic structural diagram of a system according to an embodiment of the present invention;

FIG. 14 is a schematic structural diagram of a system according to an embodiment of the present invention.

detailed description

The present invention will be further described in detail with reference to the accompanying drawings, in which FIG. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present invention without creative efforts are within the scope of the present invention.

The embodiment of the invention provides a picture management method, which can be applied to devices such as servers. As shown in FIG. 1, the method includes steps 101 to 103.

In step 101, the set picture classification feature is acquired.

In the embodiment of the present invention, the image classification feature is used to classify the image to be managed, and may have various representations. In the embodiment of the present invention, the classification information of the image refers to the information of the graphic image.

The following takes the picture to be managed as a picture stored in the web album as an example, and further describes the step. According to the different albums, the photo album may be a photo album of a person's photo, or may be a photo album of various categories such as landscapes and patterns. Therefore, the picture classification feature may be a face feature, or may be various graphic features such as having a triangle, or a sun, or a smiley face, and the like. Any image classification feature that can be acquired by using the graphic image recognition technology is acceptable, and those skilled in the art can know that there are many feature information that can be acquired by using the graphic image recognition technology, and it is impossible to perform the present invention. The embodiment does not limit the specific expression of the picture classification feature.

The following example illustrates an example of obtaining an image classification feature in an application scenario. It can be understood that the acquisition process of the image classification feature is different based on different application scenarios, and the process of acquiring the image classification feature does not affect. The following examples are not to be construed as limiting the application scenarios of the embodiments of the present invention. Optionally, obtaining the set picture category may include: displaying a picture in the network album; receiving an instruction for selecting a picture of the user to select the picture, extracting a picture classification feature of the picture corresponding to the instruction for selecting the picture, and using the picture as the set picture Classification features. Implemented in front The picture classification feature has been described in the example. In this embodiment, the picture classification feature that can be provided is different based on the picture selected by the user. Taking a portrait picture as an example, after the user selects a certain portrait picture, the face feature is extracted as the picture classification feature of the picture.

In the embodiment of the present invention, the process of acquiring the image classification feature information may be performed dynamically, that is, the image classification feature information may be acquired during the subsequent display of the acquired image. In the embodiment of the present invention, the set picture classification feature may also be filtered to obtain a preferred picture classification feature, for example, taking the intersection of various picture classification features, or taking the union of various picture classification features, and Based on the erroneous judgment, the erroneous picture classification feature in the contradictory picture classification feature is removed on the basis of the union, and the preferred rule is not limited in the embodiment of the present invention. Since the picture classification feature changes, the obtained picture will also change correspondingly, thereby realizing the technical purpose of dynamically recommending pictures. The implementation of the dynamic update picture classification feature provided by the embodiment of the present invention may include: if the picture classification feature of two or more pictures is extracted, the extracted picture classification feature is filtered to obtain a preferred picture classification feature, and The preferred picture classification feature serves as a set picture classification feature.

In step 102, the picture to be managed is searched using the picture classification feature to obtain a picture having the picture classification feature.

In the embodiment of the present invention, the picture to be managed may be a plurality of pictures in the database, and various sets are obtained in various rules, and the pictures may be stored in an album in the database, which is not limited in the embodiment of the present invention. In the embodiment of the present invention, when the picture classification feature has been obtained before the retrieval is performed, it is only necessary to automatically classify the picture. The process of classification can be done on the fly, or there can be pre-operations to achieve acceleration. The solution that is completed in real time may include: after obtaining the image classification feature, extracting the image classification characteristic data of each picture, and then performing comparison to determine whether the picture belongs to a required category; the pre-operation scheme may include: adopting a pre-processing manner to the database In the image preprocessing, the image classification feature data of all the pictures is obtained, and after the image classification feature is obtained, the image classification feature can be directly used for searching in the database, thereby saving the time of feature extraction. There may be many specific ways of searching, which are not limited in the embodiment of the present invention. The following are examples of several implementation manners, as follows: pre-storing the image classification feature data of the picture to be managed in the database; Searching for a picture in the album to be managed, and obtaining a picture having the image classification feature includes: searching in the database using the image classification feature to obtain the above Image classification feature matching image. The image classification feature data mentioned above is the data parameter value of the image classification feature mentioned above, and the parameter values of different image classification feature data based on the image classification feature will also have different representation forms, and no longer one by one. Description.

In the embodiment of the present invention, the feature data in the database may be centralized pre-processing, or may be pre-processed when the user adds a picture to a database, such as an album. An example of an implementation scheme is as follows: the following is as follows: before acquiring the set image classification feature, the method further includes: after receiving the instruction for storing the image, extracting image classification feature data of the image corresponding to the instruction for storing the image and Stored in the database.

In the embodiment of the present invention, based on a specific application scenario of a person's photo, the image classification feature may be a facial feature, and for different people, different facial features may be used, and then the facial feature is classified as a picture. The feature can automatically filter the image with the face from a large number of pictures, and further combine the pre-processing scheme to speed up the classification and reduce the waiting time of the user, as follows: the above picture classification feature is a face feature, the above The image classification feature data pre-stored in the database for the picture in the album to be managed includes: performing face recognition on the pictures in each album in the database, calculating and storing the face features of the pictures of the face and the feature similarity to the database.

Face recognition is also commonly referred to as portrait recognition and face recognition. The main step of face recognition is face detection. In actual application, face detection is mainly used for pre-processing of face recognition, that is, the position and size of the face are accurately calibrated in the image. The pattern features contained in the face image are very rich, such as histogram features, color features, template features, structural features, and Haar-like (a feature description operator) feature. Face detection is to pick out the useful information and use these features to achieve face detection. The commonly used face detection method uses the Adaboost learning algorithm based on the above features, and the Adaboost algorithm is a method for classification. The various pattern features extracted during the face detection process can be stored as parameters of the face features.

In the embodiment of the present invention, based on the foregoing facial features as the image classification feature, the embodiment of the present invention further provides a specific implementation scheme for performing retrieval based on the facial features, which is determined by the similarity in the present embodiment. It can be known by the skilled person that face clustering based on face clustering or a more complicated face recognition scheme can also be classified by facial features. The following example of using a similarity scheme as a preferred solution should not be construed as implementing the present invention. The only limitation of the example. When using similarity for graphing In the slice classification, the image classification feature is used in the above database to perform the retrieval, and obtaining the image matching the image classification feature includes: using the set face feature to retrieve the image with the feature similarity within a predetermined range in the database, and obtaining A picture whose similarity to the set face feature is within a predetermined range.

In step 103, the pictures obtained above are grouped into one category.

In the embodiment of the present invention, when the image to be managed is stored in the album, after the image classification feature is acquired, the image in the album to be managed is retrieved, and the image having the image classification feature is obtained, and the obtained image is obtained. Displayed to the user, which realizes the function of automatic recommendation, which can eliminate the manual workload of the user and reduce the complexity of the user operation, thereby improving the efficiency of the album management.

The following embodiment will take a character-based web album as an example for detailed description. At present, the album based on the character is generally marked (also called mark) function: the function refers to the user's identity information for a face (that is, which person corresponds to this face). In the embodiment of the present invention, the manner of labeling or marking may be manually selected by the user, or may be selected or confirmed by the user based on the recommendation of the system.

The embodiment of the invention proposes a scheme capable of helping the web album user to classify and centrally display the photos in the web album according to different characters appearing in the photo. In the process of implementing the embodiment of the present invention, the face detection may be performed first, and then the user is guided to mark the identity of the person corresponding to the face in the photo; then the system performs the face recognition calculation in the background by using the information marked by the user (including feature extraction and Similarity calculation), and then based on the result of face recognition, recommend to the user all the photos of the face of the character (having the face features of the above characters and all the pictures whose similarities are within a predetermined range) Guide users to confirm the identity information of these faces in batches. Based on the user's confirmation feedback, the system can also update the recommendation information in real time.

With the solution of the embodiment of the present invention, since the system can automatically recommend similar faces, the user only needs to perform a few steps to complete the tag confirmation work of the same person in the photo. In this way, the user can quickly classify different characters in his album; the web album can also classify and display images according to different characters based on the user's annotation information.

The embodiment of the invention provides another method for managing pictures, which can be applied to terminals, such as personal computers, mobile terminals and the like. As shown in FIG. 2, the method may include steps 201 to 203.

Taking a picture stored in a portrait album as an example, and using the character face as a theme, the user is provided with a way to aggregate and manage the picture by the character dimension.

In step 201, a photo with a person's face in the user's album is extracted and displayed by the face detection technology.

In step 202, the receiving user labels the displayed photos, and the displayed photos may include unlabeled photos or may include the already labeled photos, and the unlabeled photos may be regarded as photos of unknown faces.

Since the photo displayed in step 201 can have many sheets, it is possible for the user to mark a plurality of photos.

In step 203, the system uses the space album to aggregate the photos of the same face for the user according to the result of the user's annotation.

It should be noted that the photo of the same face displayed in this step is recommended by the system to the user, and may be misidentified. For example, in the similarity scheme, the range of similarity is set to be larger. Misidentification may occur. If the setting is less, the recommended situation may be missed. The specific configuration may be set by the technician according to the actual application requirements.

In addition, it should be noted that, in step 202, the user may continuously mark the photo, the system may also calculate the face data in real time and dynamically according to the marked result, and extract a face with high similarity from the unknown face, and recommend Label users to further reduce the barriers to application.

The solution of the embodiment of the invention helps the user to quickly complete the process of photo classification by using techniques such as face detection, face similarity calculation, and face automatic recommendation. The solution of the embodiment of the present invention first performs face detection, and then guides the user to mark the identity of the person corresponding to the face in the photo. The system uses the user's labeled identity information to perform face recognition calculation (including feature extraction and similarity calculation) in the background, and then based on face recognition, so as to recommend all photos of the face of the character to the user, and guide the user. Confirm the identity information of these faces in batches. The system can also update the recommendation information in real time according to the user's confirmation feedback. Since the system automatically recommends similar faces, the user only needs to perform a few steps. Mark the work of the same person in the photo. In this way, the user can quickly classify different characters in his album; the web album can also classify and display photos according to different characters based on the user's annotation information.

In order to implement the portrait classification management album function based on user annotation and system recommendation, the embodiment of the present invention mainly needs to design and implement a user annotation and system recommendation system. Based on the annotation and recommendation system, the user can complete the annotation process of the portrait efficiently and conveniently. Then, the system can automatically display and manage the corresponding portrait photos according to the user's annotation information. As shown in Figure 3, it is the full picture of the system, including the following parts:

Web Album: There are various photos.

Pre-processing system: pre-processing photos in the web album, mainly including face recognition, face feature extraction, similarity calculation, etc., which will be described in more detail in the following embodiments.

Labeling and recommendation system: It can receive the user's annotation of the photo of the web album, and recommend photos according to the user's needs according to the user's annotation. A detailed implementation of the annotation and recommendation system will be described in more detail in subsequent embodiments.

The person annotation information is information used by the annotation and recommendation system to classify the album according to the user annotation.

Classified management and display photos: Realize the display of photos recommended by the recommendation system, accept the user's selection confirmation and implement the classification management function.

As shown in FIG. 4, the preprocessing system can include the following parts:

Photo storage server: is a server that stores photos, in which the photo can be detected, that is, detecting whether there is a face in the photo, and sending a photo of the face to the task distribution server, and submitting a pre-processing task instruction;

The task distribution server is a server for the overall management of the pre-processing tasks, and distributes the pre-processing tasks to the pre-processing servers, and sends the received photos to the pre-processing servers;

Pre-processing server: Calculate the received photos, obtain facial features and feature similarity, and then store them in the face feature database and the face feature similarity database. Since the amount of calculation may be large, there may be more than one pre-processing server.

As shown in Figure 5, the annotation and recommendation system can include the following parts:

Front page: Display the photo to the user and receive the user's annotation of the photo. Sending the labeling result to the task distribution server after receiving the user's annotation of the photo;

The task distribution server: receives the user annotation and the information confirmed by the user, and distributes the information to the annotation server as a task;

Label server: label the photo and store the face label information to the database of face label data;

Recommended server: Read face tag data and face feature acquaintance information, calculate recommendation information, and send it to the front page. That is: get the result that can be used as a recommended photo to the user and send it to the front page.

The foreground page will display the recommended photo at this time, waiting for the user to confirm, and if the user has confirmed, send the confirmation message to the task distribution server.

The embodiment of the invention can help the users of the web album to classify and centrally display the photos in the web album according to different characters appearing in the photo more efficiently and quickly. Since the system can automatically recommend photos with similar facial features to the user, the user only needs to perform a few steps to complete the tag confirmation work of the same person in the photo. In this way, the user can quickly classify different characters in his album; the web album can also classify and display photos according to different characters based on the user's annotation information.

In this embodiment, the problem to be solved is mainly to help the users of the web album to classify the photos according to different people more efficiently and quickly. The technical solution proposed in this embodiment is based on user annotation and system recommendation. The system of this program can achieve this function as long as it can implement automatic or semi-automatic mode to help users sort photos according to different portraits. For example, the recommendations in the annotation and recommendation system shown in FIG. 5 can be modified based on face clustering or complex face recognition rather than simple face similarity recommendation. In addition, the pre-processing system of Figure 4 is for acceleration and is not necessarily a solution.

The embodiment of the invention further provides a picture management device, as shown in FIG. 6, comprising:

a feature acquiring unit 601, configured to acquire a set image classification feature;

The image retrieving unit 602 is configured to perform a search on the image to be managed by using the above-mentioned classification feature, and obtain a picture having the image classification feature acquired by the feature acquiring unit 601;

The display unit 603 is configured to display the obtained image by the image retrieval unit 602.

The picture determining unit 604 is configured to receive a selection instruction for selecting the displayed picture, and use the picture corresponding to the selection instruction as the picture that has been confirmed to be selected.

In the above embodiment, after the image classification feature is acquired, the image to be managed is searched, and the image having the image classification feature is obtained, and the images are classified into one class. When the picture to be managed is stored in the photo album, the picture management device can display the obtained picture to the user, thereby realizing the function of automatic recommendation, which can eliminate the manual manual workload and reduce the user's manual workload without the user manually selecting from a large number of pictures. The user operation complexity can improve the efficiency of album management.

In the implementation of the present invention, the picture classification feature is the feature information used to classify the picture to be managed, and it can have various expression forms, for example, the name marked by the user, and the name of the label is associated with a picture, this The feature information of the picture can be used as the image classification feature; or the user selects a photo, then the face feature of the photo can also be used as the image classification feature. In addition, when the to-be-processed picture is stored in the album, the album may be a photo album of a person's photo, or may be a photo album of various categories such as a landscape or a pattern, so the image classification feature may be a facial feature, It can be various graphic features such as having a triangle, or a sun, or a smiley face, and the like. As long as any classification feature that can be acquired by using the graphic image recognition technology is acceptable, those skilled in the art can know that there are many feature information that can be acquired by using the graphic image recognition technology, and it is impossible to carry out the exhaustive. The specific form of expression of the classification feature is not limited.

The following embodiment illustrates an example of obtaining an image classification feature in an application scenario. It can be understood that the acquisition process of the image classification feature is different based on different application scenarios, and the process of acquiring the classification feature does not affect the present process. The following examples are not to be construed as limiting the application scenarios of the embodiments of the present invention.

In the embodiment of the present invention, the display unit 603 is further configured to display a picture to be managed.

In the embodiment of the present invention, the feature acquiring unit 601 is configured to receive an instruction for selecting a picture of the picture by the user, and extract a classification feature of the picture corresponding to the instruction for selecting the picture, and serve as the set picture classification feature.

In the embodiment of the present invention, the process of acquiring the image classification feature information may be performed dynamically, that is, during the subsequent display of the acquired image, or before, the image classification feature information may be acquired. In the embodiment of the present invention, the set picture classification feature may also be filtered to obtain a preferred picture classification feature, for example, taking the intersection of various picture classification features, or taking various picture classification features. The union of the present invention may also be based on the erroneous judgment, and the erroneous image classification feature in the contradictory picture classification feature is removed on the basis of the union, and the preferred rule is not limited in the embodiment of the present invention. Since the picture classification feature changes, the obtained picture will also change correspondingly, thereby realizing the technical purpose of dynamically recommending pictures. The implementation of the dynamic update picture classification feature provided by the embodiment of the present invention may include: as shown in FIG. 7, the device further includes: an update unit 701, configured to: if the feature acquisition unit 601 extracts two or more pictures For the picture classification feature, the extracted picture classification features are filtered to obtain a preferred picture classification feature, and the preferred picture classification feature is used as the set picture classification feature.

In the embodiment of the present invention, the picture to be managed may be a plurality of pictures in the database, and various sets are obtained in various rules, and the pictures may be stored in an album in the database, which is not limited in the embodiment of the present invention. In the embodiment of the present invention, when the picture classification feature has been obtained before the retrieval is performed, it is only necessary to automatically classify the picture. The process of classification can be done on the fly, or there can be pre-operations to achieve acceleration. The solution that is completed in real time may include: after obtaining the image classification feature, extracting the image classification characteristic data of each picture, and then performing comparison to determine whether the picture belongs to a required category; the pre-operation scheme may include: adopting a pre-processing manner to the database In the image preprocessing, the image classification feature data of all the pictures is obtained, and after the image classification feature is obtained, the image classification feature can be directly used for searching in the database, thereby saving the time of feature extraction. There may be many specific ways of searching, which are not limited by the embodiment of the present invention. Examples of several implementation manners are given below, as follows:

Pre-storing image classification feature data of the above-mentioned picture to be managed in the database;

The picture retrieval unit 602 searches the database using the picture classification feature to obtain a picture that matches the picture classification feature.

In the embodiment of the present invention, the feature data in the database may be centralized pre-processing, or may be pre-processed when the user adds a picture to a database, such as an album. An example of an implementation is given below, as follows: As shown in FIG. 8, the foregoing apparatus further includes:

The feature extraction unit 801 is configured to extract image classification feature data of the image in the album to be managed and store the data in the database before acquiring the set image classification feature.

In the embodiment of the present invention, based on a specific application scenario of a person's photo, the image classification feature may be a facial feature, and for different people, different facial features may be used, then the facial feature is As the image classification feature, it is possible to automatically filter the image with the face from a large number of pictures, and further combining the pre-processing scheme can speed up the classification and reduce the waiting time of the user, as follows: the above picture classification feature is a face feature As shown in FIG. 9, the device further includes: a face recognition unit 901, configured to perform face recognition on a picture in each album in the database, calculate and store a face feature of the face image, and feature similarity to the database. .

In the embodiment of the present invention, based on the foregoing facial features as the image classification feature, the embodiment of the present invention further provides a specific implementation scheme for performing retrieval based on the facial features, which is determined by the similarity in the present embodiment. It can be known by the skilled person that face clustering based on face clustering or a more complicated face recognition scheme can also be classified by facial features. The following example of using a similarity scheme as a preferred solution should not be construed as implementing the present invention. The only limitation of the example. When the image classification is performed by using the similarity degree, the image retrieving unit 602 is configured to search for a picture whose feature similarity is within a predetermined range in the database by using the set facial feature, and obtain the similarity with the set facial feature. A picture within the predetermined range.

An embodiment of the present invention provides another picture management apparatus. As shown in FIG. 10, the apparatus includes one or more processors 1001, and a storage medium 1002 storing operation instructions. When operating an operation instruction in the storage medium 1002, The processor 1001 can perform the following steps:

Obtaining the set image classification feature;

Using the above picture classification feature, the picture to be managed is searched to obtain a picture having the above picture classification feature;

The pictures obtained above are grouped into one category.

In the embodiment of the present invention, when the operation instruction in the storage medium 1001 is run, the processor 1002 may further perform the following steps:

Receiving an instruction for the user to select a picture, extracting a picture classification feature of the picture corresponding to the instruction, and using the picture classification feature as the setting.

If the picture classification features of two or more pictures are extracted, the extracted picture classification features are filtered to obtain a preferred picture classification feature, and the preferred picture classification feature is set as Image classification feature.

Another embodiment of the present invention further provides a device, as shown in FIG. 11, comprising: a receiver 1101, a transmitter 1102, a processor 1103, a display 1104, and a memory 1105.

The processor 1103 is configured to acquire a set image classification feature, use the image classification feature to retrieve a picture to be managed, obtain a picture having the picture classification feature, display the obtained picture, and receive the displayed picture. A selection instruction is selected, and the picture corresponding to the selection instruction is used as the picture that has been confirmed to be selected.

In the embodiment of the present invention, the image classification feature is feature information used for classifying the image to be managed, and may have various expression forms, for example, a name marked by the user, and the name of the annotation is associated with a certain image, The feature information of the picture can be used as the picture classification feature; or the user selects a picture, then the face feature of the picture can also be used as the picture classification feature. In addition, when the to-be-processed picture is stored in the album, the album may be a photo album of a person's photo, or may be a photo album of various categories such as a landscape or a pattern, so the image classification feature may be a facial feature, It can be various graphic features such as having a triangle, or a sun, or a smiley face, and the like. As long as any classification feature that can be acquired by using the graphic image recognition technology is acceptable, those skilled in the art can know that there are many feature information that can be acquired by using the graphic image recognition technology, and it is impossible to carry out the exhaustive. The specific form of expression of the classification feature is not limited.

In the embodiment of the present invention, optionally, the processor 1103 may acquire the set classification feature by: controlling to display a picture to be managed; and receiving an instruction for selecting a picture of the picture by the user, The picture classification feature of the picture corresponding to the instruction for selecting the picture is extracted and used as the set picture classification feature.

In the embodiment of the present invention, the process of acquiring the image classification feature information may be performed dynamically, that is, during the subsequent display of the acquired image, or before, the image classification feature information may be acquired. In the embodiment of the present invention, the set picture classification feature may also be filtered to obtain a preferred picture classification feature, for example, taking the intersection of various picture classification features, or taking the union of various picture classification features, and Based on the erroneous judgment, the erroneous picture classification feature in the contradictory picture classification feature is removed on the basis of the union, and the preferred rule is not limited in the embodiment of the present invention. Since the picture classification feature changes, the obtained picture will also change correspondingly, thereby realizing the technical purpose of dynamically recommending pictures. The implementation of the dynamic update picture classification feature provided by the embodiment of the present invention may include: if the picture classification feature of two or more pictures is extracted, the extracted picture classification feature is filtered to obtain a preferred picture classification feature, and The preferred picture classification feature serves as a set picture classification feature.

In the embodiment of the present invention, the picture to be managed may be stored in the form of an album, which is not limited in the embodiment of the present invention. In the embodiment of the present invention, when the picture classification feature has been obtained before the retrieval is performed, it is only necessary to automatically classify the picture. The process of classification can be done on the fly, or there can be pre-operations to achieve acceleration. The solution that is completed in real time may include: after obtaining the image classification feature, extracting the image classification characteristic data of each picture, and then performing comparison to determine whether the picture belongs to a required category; the pre-operation scheme may include: adopting a pre-processing manner to the database In the image preprocessing, the image classification feature data of all the pictures is obtained, and after the image classification feature is obtained, the image classification feature can be directly used for searching in the database, thereby saving the time of feature extraction. There may be many types of specific search methods, which are not limited by the embodiments of the present invention. The following are examples of several implementation manners, as follows: pre-storing the image classification features of the pictures in the album to be managed in the database. Data; the processor 1103 searches for a picture in the album to be managed, and the image having the image classification feature is obtained by using the image classification feature in the database to obtain a match with the image classification feature. picture of.

In the embodiment of the present invention, the feature data in the database may be centralized pre-processing, or may be pre-processed when the user adds a picture to a database, such as an album. An example of an implementation is given below, as follows: The processor 1103 is configured to extract the set image classification feature. The picture classification feature data of the picture in the managed album is stored and stored in the database.

In the embodiment of the present invention, based on a specific application scenario of a person's photo, the image classification feature may be a face feature, and for different people, different face features may be used, and then the face feature is used as a picture classification feature. , the image of the face can be automatically filtered from a large number of pictures, and the combination of the pre-processing scheme can speed up the classification and reduce the waiting time of the user, as follows: the above picture classification feature is a face feature, the processor 1103 may pre-store the image classification feature data of the picture in the album to be managed in the database by performing face recognition on the picture in each album in the database, calculating and storing the face feature and feature of the picture of the face Similarity to the database.

In the embodiment of the present invention, based on the foregoing facial features as the image classification feature, the embodiment of the present invention further provides a specific implementation scheme for performing the retrieval based on the facial features, which is determined by the similarity in the embodiment, and is technical in the field. It can be known that a face-based clustering or a more complicated face recognition scheme can also be classified by facial features. The following example of using a similarity scheme as a preferred scheme should not be construed as an embodiment of the present invention. The only limit. When the similarity is used for the image classification, the processor 1103 performs the above-mentioned image classification feature for searching in the database, and obtaining the image matching the image classification feature can be implemented by using the set facial feature in the database. A picture in which the feature similarity is within a predetermined range is retrieved, and a picture whose similarity with the set face feature is within a predetermined range is obtained.

The embodiment of the invention further provides a picture management system, as shown in FIG. 12, comprising: a database server 1201, an annotation and recommendation server 1202.

The database server 1201 is configured to store a to-be-processed picture;

The labeling and recommendation server 1202 is configured to receive the set image classification feature, use the received image classification feature to perform a search on the image in the database, obtain a picture having the image classification feature, and send the obtained image to the terminal. For the above terminal to make a selection.

In the above embodiment, after the image classification feature is acquired, the image to be managed is retrieved, and the image having the image classification feature is obtained, and the obtained image is displayed to the user, thereby realizing the function of automatic recommendation, and the user can be Manual selection in the picture reduces the manual workload of the user and reduces the complexity of user operations.

The following example illustrates an example of obtaining an image classification feature in an application scenario. It can be understood that the acquisition process of the image classification feature is different based on different application scenarios, and the process of acquiring the image classification feature does not affect. For the implementation of the embodiments of the present invention, the following examples should not be construed as limiting the application scenarios of the embodiments of the present invention. Optionally, the labeling and recommendation server 1202 is configured to receive an instruction for selecting a picture from a terminal, and extract an instruction corresponding to the selected picture. The classification characteristics of the picture and as a set of picture classification features.

In the embodiment of the present invention, the process of acquiring the image classification feature information may be performed dynamically, that is, during the subsequent display of the acquired image, or before, the image classification feature information may be acquired. In the embodiment of the present invention, the set picture classification feature may also be filtered to obtain a preferred picture classification feature, for example, taking the intersection of various picture classification features, or taking the union of various picture classification features, and Based on the erroneous judgment, the erroneous picture classification feature in the contradictory picture classification feature is removed on the basis of the union, and the preferred rule is not limited in the embodiment of the present invention. Since the picture classification feature changes, the obtained picture will also change correspondingly, thereby realizing the technical purpose of dynamically recommending pictures. The implementation scheme of the dynamic update picture classification feature provided by the embodiment of the present invention may include: the labeling and recommendation server 1202, and the image classification feature of the extracted image if the image classification feature of two or more pictures is extracted. The screening is performed to obtain a preferred image classification feature, and the preferred image classification feature is used as the set image classification feature.

The database server 1201 is further configured to store image classification feature data of a picture in the album;

The labeling and recommendation server 1202 is configured to perform the search using the image classification feature in the database server 1201 to obtain a picture matching the picture classification feature.

Further, in order to achieve the management of the pre-processing process, the embodiment of the present invention provides the following solution: as shown in FIG. 13, the system further includes: a pre-processing server 1301;

The pre-processing server 1301 is configured to extract the picture classification feature data of the picture to be managed, and store the extracted picture classification feature in the database server 1201.

In the embodiment of the present invention, based on a specific application scenario of a person's photo, the image classification feature may be a facial feature, and for different people, different facial features may be used, and then the facial feature is classified as a picture. The feature can automatically filter the image with the face from a large number of pictures, and further combine the pre-processing scheme to speed up the classification and reduce the waiting time of the user, as follows: the above picture classification feature is a face feature, the above pre- The processing server 1301 is configured to perform face recognition on the pictures in each album in the database, calculate face features and feature similarities of the pictures of the human face, and calculate the calculated face features and feature similarities into the database server 1201. .

Further, in order to implement the management of the pre-processing process, the embodiment of the present invention provides the following solution: as shown in FIG. 14, the system further includes: a task distribution server 1401;

The task distribution server 1401 is configured to distribute the picture of the face of each album stored in the database server 1201 to each pre-processing server 1301;

The pre-processing server 1301 is configured to perform face recognition on the received picture distributed by the task distribution server 1401, calculate face features and feature similarity of the face image, and calculate the calculated face features and features. The degree is reached in the above database server 1201.

In the embodiment of the present invention, based on the foregoing facial features as the image classification feature, the embodiment of the present invention further provides a specific implementation scheme for performing retrieval based on the facial features, which is determined by the similarity in the present embodiment. It can be known by the skilled person that face clustering based on face clustering or a more complicated face recognition scheme can also be classified by facial features. The following example of using a similarity scheme as a preferred solution should not be construed as implementing the present invention. The only limitation of the example. When the similarity is used for image classification, the pre-processing server 1301 is configured to use the set facial features in the above database. The server 1201 retrieves a picture whose feature similarity is within a predetermined range, and obtains a picture whose similarity with the set face feature is within a predetermined range.

It should be noted that, in the foregoing apparatus and terminal embodiment, each unit included is only divided according to functional logic, but is not limited to the above division, as long as the corresponding function can be implemented; The specific names are also for convenience of distinguishing from each other and are not intended to limit the scope of the present invention.

In addition, those skilled in the art can understand that all or part of the steps in implementing the foregoing method embodiments may be performed by a program to instruct related hardware, and the corresponding program may be stored in a computer readable storage medium. The storage medium may be a read only memory, a magnetic disk or an optical disk or the like.

The above is only a preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the embodiments of the present invention. All should be covered by the scope of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

A picture management method, comprising:

Obtaining the set image classification feature;

Using the picture classification feature to search for a managed picture to obtain a picture having the picture classification feature;

Classify the images obtained.
The method according to claim 1, wherein the obtaining the set picture classification feature comprises:

Receiving an instruction for the user to select a picture, extracting a picture classification feature of the picture corresponding to the instruction, and using the picture classification feature as the setting.
The method of claim 2 wherein

If the picture classification features of the two or more pictures are extracted, the extracted picture classification features are filtered to obtain a preferred picture classification feature, and the preferred picture classification feature is used as the set picture classification feature.
The method according to any one of claims 1 to 3, wherein when the picture classification feature data of the picture to be managed is pre-stored in the database, the picture to be managed is searched to obtain the picture classification Features of the image include:

The image classification feature is used in the database to perform a retrieval, and a picture matching the image classification feature is obtained.
The method according to claim 4, further comprising: before acquiring the set picture classification feature:

Extracting picture classification feature data of the picture to be managed and storing.
The method according to claim 4, wherein the picture classification feature is a face feature, and the picture classification feature data pre-stored in the database for the picture to be managed comprises:

The face to be managed is subjected to face recognition, and the face features of the face of the face and the feature similarity are calculated and stored in the database.
The method according to claim 6, wherein the searching in the database using the picture classification feature to obtain a picture matching the picture classification feature comprises:

Retrieving a graph of feature similarity within a predetermined range in the database using the set face features A slice obtains a picture whose similarity to the set face feature is within a predetermined range.
A picture management apparatus, comprising: one or more processors and a storage medium storing operation instructions, when executing an operation instruction in the storage medium, the processor performs the following steps:

Obtaining the set image classification feature;

Using the picture classification feature to search for a managed picture to obtain a picture having the picture classification feature;

Classify the images obtained.
The apparatus according to claim 8, wherein said processor further performs the following steps:

Receiving an instruction for the user to select a picture, extracting a picture classification feature of the picture corresponding to the instruction, and using the picture classification feature as the setting. The device according to claim 9, wherein the processor is further configured to perform the following steps:

If the picture classification features of the two or more pictures are extracted, the extracted picture classification features are filtered to obtain a preferred picture classification feature, and the preferred picture classification feature is used as the set picture classification feature.
A picture management system, comprising:

Database server, annotation and recommendation server,

The database server is configured to store a picture to be processed;

And the labeling and recommendation server is configured to receive the set image classification feature, use the image classification feature to retrieve the image in the database server, obtain a picture having the image classification feature, and send the obtained image to the terminal. A picture for the terminal to select.
The system of claim 11 wherein:

And the labeling and recommendation server is configured to receive an instruction for selecting a picture from the terminal, and extract a picture classification feature of the picture corresponding to the instruction, and serve as the set picture classification feature.
The system of claim 12 wherein:

The labeling and recommendation server is further configured to: if the image classification features of two or more pictures are extracted, filter the extracted picture classification features to obtain a preferred picture classification feature, and use the preferred classification feature as the Set the image classification feature.
A system according to any one of claims 11 to 13, wherein

The database server is further configured to store image classification feature data of the to-be-managed picture;

The labeling and recommendation server is configured to perform, by using the image classification feature in the database server, to obtain a picture that matches the picture classification feature.
The system of claim 14 wherein said system further comprises: a pre-processing server;

The pre-processing server is configured to extract picture classification feature data of the to-be-managed picture, and store the extracted picture classification feature into the database server.