WO2023124134A1

WO2023124134A1 - File processing method and apparatus, electronic device, computer storage medium and program

Info

Publication number: WO2023124134A1
Application number: PCT/CN2022/113914
Authority: WO
Inventors: 贺文峰; 王康; 何俊峰; 霍明明; 郭如意
Original assignee: 上海商汤智能科技有限公司
Priority date: 2021-12-28
Filing date: 2022-08-22
Publication date: 2023-07-06
Also published as: CN114329023A

Abstract

Disclosed in embodiments of the present invention are a file processing method and apparatus, a device, a computer storage medium and a computer program. The method comprises: obtaining an image feature set of a target object; determining a similarity set between each image feature in the image feature set and other image features in the image feature set; determining a representative feature of the image feature set according to the similarity set; and determining an image corresponding to the representative feature as a cover image of a personnel image file corresponding to the target object.

Description

File processing method, device, electronic device, computer storage medium and program

Cross References to Related Applications

This application is based on the Chinese patent application with the application number 202111626946.5, the filing date is December 28, 2021, and the title is "file processing method, device, electronic equipment and computer storage medium", and claims the priority of the Chinese patent application, The entire content of this Chinese patent application is hereby incorporated by reference into this application.

technical field

The present disclosure relates to computer vision technology, and relates to but not limited to a file processing method, device, electronic equipment, computer storage medium and computer program.

Background technique

At present, in the process of building a smart city, a series of computer vision technologies such as target detection, feature extraction, attribute detection, and face clustering can be used to classify and label camera capture data. Based on these data, Personnel image files can be defined, that is, the trajectory of each person can be defined.

Contents of the invention

Embodiments of the present disclosure provide a file processing method, device, electronic equipment, computer storage medium and computer program.

An embodiment of the present disclosure provides a file processing method, the method comprising:

Acquire the image feature set of the target object; determine the similarity set between each image feature in the image feature set and other image features in the image feature set; determine the image feature set according to the similarity set Representative feature: determining the image corresponding to the representative feature as the cover image of the file corresponding to the target object.

In some embodiments of the present disclosure, the determining the representative feature of the image feature set according to the similarity set includes: determining the local density of each image feature according to the similarity set; The similarity set and the local density of each image feature determine the density point interval of each image feature; according to the local density and the density point interval of each image feature, determine the image feature Representative characteristics of the collection.

It can be seen that the local density and density point interval of each image feature can objectively and accurately reflect the distribution of image features in the feature space, so that the image corresponding to the representative feature is a representative image selected objectively and accurately, that is, The embodiment of the present disclosure can accurately select a representative cover image.

In some embodiments of the present disclosure, the determining the local density of each image feature according to the similarity set includes: determining the similarity with each image feature in the image feature set At least one image feature greater than a threshold; using the sum of similarities between the at least one image feature and each image feature as the local density of each image feature.

It can be understood that after the image feature set is obtained, the features of the image feature set have a certain feature distribution in the feature space. In some areas, the feature distribution will be relatively concentrated, that is, the local density of the feature is high, indicating that the feature area has Representativeness, find the center point of the density region as the representative feature of the region; therefore, in the embodiments of the present disclosure, by calculating the local density of each image feature, it is beneficial to determine the representative feature of the image feature set.

In some embodiments of the present disclosure, the determining the density point interval of each image feature according to the similarity set and the local density of each image feature includes: the first in the image feature set When the image feature is not the image feature with the largest local density in the image feature set, select an image feature with a local density greater than the local density of the first image feature in the image feature set, and the selected image feature The minimum value of the similarity with the first image feature is used as the density point interval of the first image feature; the first image feature is any image feature in the image feature set;

In the case of obtaining the density point spacing of each image feature in the image feature set except for the image feature with the highest local density, each image feature in the set of image features except the image feature with the highest local density In the density point interval of the image feature, the minimum value of the density point interval is used as the density point interval of the image feature with the highest local density in the image feature set.

It is understandable that there may be more than one image feature with high local density. In order to reduce the interaction between different image features with high local density, it is necessary to determine the image with high local density and other high local density in the feature space. An image feature with a sufficient similarity distance is used as a representative feature; here, the similarity distance between two image feature points is negatively correlated with the similarity between two image features; The point interval characterizes the similarity between image features with high local density and other image features with high local density; by determining the density point interval of image features, it is helpful to accurately determine the image features with high local density and other high local density in the feature space. Image features with a sufficient similarity distance are conducive to accurately determine representative features.

In some embodiments of the present disclosure, the determining a representative feature of the image feature set according to the local density and the density point interval of each image feature includes: according to each image feature The local density and the density point interval of , obtain the representative score of each image feature, the representative score of each image feature is positively correlated with the local density, and proportional to the density point interval Negative correlation; in the image feature set, at least one feature whose representative score is greater than or equal to a representative score threshold is selected as the representative feature.

It can be seen that when the local density of an image feature is high or the density point interval is low, the representative score of the image feature will be higher, and the image feature will be more likely to become a representative feature; that is, there It is beneficial to use a feature with high local density and a small similarity with other high local density features as a representative feature, so that the embodiment of the present disclosure can more accurately determine the representative feature in the image feature set.

In some embodiments of the present disclosure, the deriving the representative score of each image feature according to the local density and the density point interval of each image feature includes: The ratio of the local density of the image feature to the density point interval is used as the representative score of each image feature. It can be seen that in the embodiments of the present disclosure, the representative score of each image feature can be relatively easily obtained by calculating the ratio of the local density of each image feature to the density point interval.

In some embodiments of the present disclosure, after the cover image of the archive is determined for the first time, the method further includes: in response to an update event of the cover image, implementing the following steps to implement the cover image of the archive Update: obtaining a set of image features of the target object, determining a representative feature of the set of image features, and determining a cover image of the profile; wherein:

The update event of the cover image includes at least one of the following: when the file is updated; when an instruction to update the cover image is acquired; when the cover image is periodically updated.

It can be seen that in the embodiment of the present disclosure, when it is determined to update the file, or when it is determined that the cover image needs to be updated, the cover image can be refreshed by repeatedly executing the file processing method of the embodiment of the present disclosure; when the cover image is periodically updated In some cases, by repeatedly executing the file processing method of the embodiment of the present disclosure, the continuous update of the cover image can be realized. Since there is no need to manually correct the cover image, the operation and maintenance cost can be reduced; in some embodiments, when updating the file In some cases, the image in the file is a recently acquired image, that is, the embodiment of the present disclosure can select the cover image based on the recently acquired image, so that the selected cover image can reflect the recent state of the target object, thereby facilitating subsequent targeting of the target object. Perform a feature search.

In some embodiments of the present disclosure, the acquisition of the image feature set of the target object includes: determining the acquisition time and the current time duration of each image in the file corresponding to the target object; in the file, the The image whose duration exceeds the life cycle of the image is filtered out to obtain an updated file; feature extraction is performed on the updated image file to obtain an image feature set of the target object.

In the embodiment of the present disclosure, the images in the file whose collection time and current time are longer than the life cycle of the image are filtered out, which can improve the reliability of the process of selecting the cover image; in some embodiments, the embodiment of the present disclosure can A representative image is selected based on the most recently collected image, so that the selected representative image can reflect the latest state of the target object.

In some embodiments of the present disclosure, prior to updating the cover image, the method further includes: in response to adding a new image of the target object to the profile, determining the number of images in the profile; in response to When the number of images in the archive exceeds the preset number, the images in the archive are sequentially filtered out according to the order of collection time in the archive from early to late until the number of images in the archive is less than or equal to the preset amount.

Understandably, in the case of adding a new image of the target object to the archive, in response to the number of images in the archive exceeding a preset number, it is indicated that the images in the archive consume more disk resources, in this case , by filtering out the images in the archive, it is beneficial to reduce the disk resources occupied by the images in the archive.

An embodiment of the present disclosure also proposes a file processing device, which includes an acquisition part, a first processing part, a second processing part and a third processing part, wherein,

The acquisition part is configured to acquire the image feature set of the target object;

A first processing part configured to determine a similarity set between each image feature in the set of image features and other image features in the set of image features;

The second processing part is configured to determine representative features of the image feature set according to the similarity set;

The third processing part is configured to determine the image corresponding to the representative feature as the cover image of the file corresponding to the target object.

An embodiment of the present disclosure also provides an electronic device, including a processor and a memory for storing a computer program that can run on the processor; wherein, the processor is used to run the computer program to execute any one of the above-mentioned files Approach.

An embodiment of the present disclosure also provides a computer storage medium, on which a computer program is stored, and when the computer program is executed by a processor, any one of the file processing methods above is implemented.

An embodiment of the present disclosure also provides a computer program, the computer program includes computer readable code, when the computer readable code is run in the electronic device, the processor in the electronic device executes any one of the above files Approach.

It can be seen that the embodiments of the present disclosure can obtain the representative features in the image feature set according to the similarity set between each image feature in the image feature set and other image features. Therefore, the image corresponding to the representative feature is The representative image is accurately determined according to the similarity between image features, that is, the embodiment of the present disclosure can accurately select a representative cover image from the archive.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.

Description of drawings

The accompanying drawings here are incorporated into the description and constitute a part of the present description. These drawings show embodiments consistent with the present disclosure, and are used together with the description to explain the technical solution of the present disclosure.

FIG. 1 is a flowchart of a file processing method according to an embodiment of the present disclosure;

FIG. 2 is a flow chart of determining representative features in an embodiment of the present disclosure;

Fig. 3 is a flowchart of determining the local density of image features in an embodiment of the present disclosure;

Fig. 4 is a flowchart of determining the density point interval of image features in an embodiment of the present disclosure;

FIG. 5 is a schematic diagram of the principle of determining representative features in an embodiment of the present disclosure;

FIG. 6 is another flow chart for determining representative features in an embodiment of the present disclosure;

FIG. 7 is a flow chart of acquiring an image feature set of a target object in an embodiment of the present disclosure;

FIG. 8 is a flow chart of filtering images in archives in an embodiment of the present disclosure;

FIG. 9 is a schematic structural diagram of a file processing device according to an embodiment of the present disclosure;

FIG. 10 is a schematic structural diagram of an electronic device according to an embodiment of the present disclosure.

Detailed ways

In related technologies, computer vision technology can be used to process the captured images, so as to obtain the files of each person. Based on the files of each person, data analysis tasks such as target person retrieval and track query can be well realized. However, the file itself is a kind of relational information, which cannot be displayed intuitively. In addition, for massive snapshot images, it is very difficult to achieve full retrieval, but in actual scenarios, only some typical snapshot images under each file need to be retrieved. Therefore, it is necessary to be able to select the most representative image in the archive as the cover image of the archive.

The traditional cover image is an arbitrary snapshot image randomly selected in the archives. This method has the following two problems: 1) The quality of the selected image cannot be controlled, and the selected image may have a bad shooting angle, blurred image, inappropriate zoom ratio, etc. 2) The randomly selected image may be a historical image that has expired and been deleted, resulting in the loss of the cover image. Moreover, if the capture time of the randomly selected image is longer than the current time, the image cannot represent Recent status of the archivist.

In view of the above technical problems, the technical solutions of the embodiments of the present disclosure are proposed.

The present disclosure will be described in further detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the embodiments provided here are only used to explain the present disclosure, not to limit the present disclosure. In addition, the embodiments provided below are some embodiments for implementing the present disclosure, rather than providing all the embodiments for implementing the present disclosure. In the case of no conflict, the technical solutions recorded in the embodiments of the present disclosure can be combined in any manner implement.

It should be noted that in the embodiments of the present disclosure, the term "comprises", "comprises" or any other variation thereof is intended to cover a non-exclusive inclusion, so that a method or device comprising a series of elements not only includes the explicitly stated elements, but also include other elements not explicitly listed, or also include elements inherent in implementing the method or apparatus. Without further limitations, an element defined by the phrase "comprising a..." does not exclude the presence of additional related elements (such as steps in the method or A unit in an apparatus, for example, a unit may be part of a circuit, part of a processor, part of a program or software, etc.).

For example, the file processing method provided by the embodiment of the present disclosure includes a series of steps, but the file processing method provided by the embodiment of the present disclosure is not limited to the steps described, and similarly, the device provided by the embodiment of the present disclosure is not limited to include the specified steps. The recorded part may also include the part that needs to be set for obtaining relevant information or processing based on the information.

Embodiments of the present disclosure may be applied to electronic devices such as terminals and servers. Here, a terminal can be a thin client, a thick client, a handheld or laptop device, a microprocessor-based system, a set-top box, programmable consumer electronics, a network personal computer, a small computer system, etc., and a server can be a server computer Systems Small computer systems, mainframe computer systems, and distributed cloud computing technology environments including any of the above systems, etc.

Electronic devices such as terminals and servers may include program modules for executing instructions. Generally, program modules may include routines, programs, objects, components, logic, data structures, etc., that perform particular tasks. The computer system/server can be practiced in distributed cloud computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed cloud computing environment, program modules may be located in both local and remote computing system storage media including storage devices.

The embodiment of the present disclosure proposes a file processing method, which can be applied to intelligent video analysis, smart city or other image analysis scenarios.

Fig. 1 is a flowchart of the file processing method of the embodiment of the present disclosure, as shown in Fig. 1, the process may include:

Step 101: Obtain an image feature set of a target object.

In the embodiment of the present disclosure, the image feature set of the target object is obtained by extracting the image feature from the file. Here, an archive represents a collection of images of a target object. For example, the target object can be a person, a vehicle, or other types of objects, and the images in the archive can be images collected by image acquisition equipment; the format of the images in the archive can be a joint image Joint Photographic Experts Group (JPEG), Bitmap (Bitmap, BMP), Portable Network Graphics (PNG) or other formats; it should be noted that this is only for the format and source of the images in the file An example is used for illustration, and the embodiment of the present disclosure does not limit the format and source of the image in the file.

Exemplarily, after acquiring the image collection of the target object, a pre-trained feature extraction network can be used to perform feature extraction on the images in the image collection to obtain the image feature collection of the target object. In the embodiment of the present disclosure, the feature extraction network does not Network structure and types are defined.

Exemplarily, after obtaining the image feature sets of at least two objects, the image feature sets of the target object can be obtained by clustering the image feature sets of the at least two objects; Exemplarily, clustering the image feature sets The class method can be a partition-based clustering algorithm, a hierarchical clustering algorithm, or a density-based clustering algorithm.

Exemplarily, in the case where the target object is a person, the image features in the image feature set may be facial features, human body features, etc.; in the case of a motor vehicle, the image features in the image feature set may be machine Motor vehicle features; when the target object is a non-motor vehicle, the image features in the image feature set may be non-motor vehicle features.

Step 102: Determine a similarity set between each image feature in the image feature set and other image features in the image feature set.

In one implementation, let C _i represent the image feature set of the target object, _i is an integer greater than or equal to 1; for the jth feature x _ij in the image feature set C _i , it can be determined that The similarity between the jth feature x _ij and other features, where j is an integer greater than or equal to 1. After determining the similarity between the jth feature x _ij and other features in the image feature set _Ci , by traversing the value of j, the relationship between each image feature in the image feature set and other image features in the image feature set can be determined collection of similarities.

Step 103: Determine representative features of the image feature set according to the similarity set.

Step 104: Determine the image corresponding to the representative feature as the cover image of the file corresponding to the target object.

In the embodiment of the present disclosure, since the image features in the image feature set are extracted from the image of the target object, the image corresponding to each image feature in the image feature set can be determined, so that after determining the representative feature, it can be An image corresponding to a representative feature is identified in a profile corresponding to the target object.

In practical applications, the above step 101 to step 104 can be implemented based on the processor of the electronic device, and the above processor can be an application specific integrated circuit (Application Specific Integrated Circuit, ASIC), a digital signal processor (Digital Signal Processor, DSP), Digital Signal Processing Device (DSPD), Programmable Logic Device (PLD), Field Programmable Gate Array (Field Programmable Gate Array, FPGA), Central Processing Unit (Central Processing Unit, CPU), At least one of controller, microcontroller, microprocessor.

It can be seen that the embodiments of the present disclosure can obtain the representative features in the image feature set according to the similarity set between each image feature in the image feature set and other image features. Therefore, the image corresponding to the representative feature is The representative image is accurately determined according to the similarity between image features, that is, the embodiment of the present disclosure can accurately select a representative cover image from the archive. In some embodiments, when the representative image is used as the cover image of the archive, compared with randomly selected images, the cover image selected by the embodiments of the present disclosure has certain interpretability and typicality, and can better represent Files are better readable to the human eye.

In some embodiments of the present disclosure, referring to FIG. 2 , according to the similarity set, the process of determining the representative feature of the image feature set may include:

Step 1031: Determine the local density of each image feature according to the similarity set.

Exemplarily, the local density of the image feature is used to represent the density of the local area of the image feature in the feature space.

Referring to FIG. 3, according to the similarity set, the process of determining the local density of each image feature may include:

Step 10311: In the image feature set, determine at least one image feature whose similarity with each image feature is greater than a threshold.

Step 10312: Take the sum of the similarities between at least one image feature and each image feature as the local density of each image feature.

In one implementation, when C _i represents the image feature set of the target object, for the j-th feature x _ij in the feature set C _i , it is possible to search for the j-th feature x _ij in the image feature set C _i The set A of features whose similarity is greater than the set similarity threshold, the sum of the similarities between each feature in set A and the jth feature x _ij is taken as the local density α _ij of the jth feature x _ij .

Step 1032: Determine the density point interval of each image feature according to the similarity set and the local density of each image feature.

Referring to Figure 4, according to the similarity set and the local density of each image feature, the process of determining the density point interval of each image feature may include:

Step 10321: When the first image feature in the image feature set is not the image feature with the largest local density in the image feature set, select an image feature with a local density greater than that of the first image feature in the image feature set, and select The minimum value of the similarity between the image feature and the first image feature is used as the density point interval of the first image feature; the first image feature is any image feature in the image feature set.

In one implementation, for the jth feature x _ij in the image feature set C _i above, if the jth feature x _ij is not the image feature with the highest local density in the image feature set C _i , it can be in the image Select the image feature whose local density is greater than the local density of the jth feature x _ij in the feature set C _i , and use the minimum value β _ij of the similarity between the selected image feature and the jth feature x _ij as the value of the jth feature x _ij Density point spacing.

Step 10322: In the case of obtaining the density point interval of each image feature in the image feature set except the image feature with the highest local density, the density points of each image feature in the image feature set except the image feature with the highest local density In the interval, the minimum value of the density point interval is used as the density point interval of the image feature with the highest local density in the image feature set.

It is understandable that there may be more than one image feature with high local density. In order to reduce the interaction between different image features with high local density, it is necessary to determine the image with high local density and other high local density in the feature space. An image feature with a sufficient similarity distance is used as a representative feature; here, the similarity distance between two image feature points is negatively correlated with the similarity between two image features; The point interval characterizes the similarity between the image feature with high local density and other image features with high local density; referring to Figure 5, the solid circle represents the image feature, and the two image features pointed by the straight line have higher local density, that is, There are many image features nearby; and, the similarity distance between the first image feature 501 and the second image feature 502 is relatively large, and the similarity between the first image feature 501 and the second image feature 502 is small. In an example, the density point interval of the second image feature 502 is: the similarity between the first image feature 501 and the second image feature 502 .

Step 1033: Determine representative features of the image feature set according to the local density and density point interval of each image feature.

In some embodiments of the present disclosure, referring to FIG. 6, the implementation of step 1033 may include:

Step 10331: Obtain the representative score of each image feature according to the local density and the density point interval of each image feature. The representative score of each image feature is positively correlated with the local density and negatively correlated with the density point interval.

Step 10332: In the image feature set, select at least one feature whose representative score is greater than or equal to the representative score threshold as a representative feature.

In the embodiment of the present disclosure, the representative score threshold may be set according to actual requirements. In one implementation, the image features in the image feature set can be sorted in the order of representative scores from high to low, and the representative score of the kth feature in the sorting result can be used as the representative score threshold, namely , select the top k features in the sorting results as representative features.

In some embodiments of the present disclosure, the ratio of the local density of each image feature to the density point interval may be used as the representative score of each image feature.

In an implementation manner, for the jth feature x _ij in the feature set C _i above, the representative score of the jth feature x _ij may be determined as α _ij /β _ij . Then, at least one feature with a representative score greater than or equal to a representative score threshold may be selected from the above feature set C _i .

In some embodiments of the present disclosure, after determining the cover image of the archive for the first time, in response to the update event of the cover image, the following steps may be performed to update the cover image of the archive: acquiring the image feature set of the target object, determining Representative features of the set of image features, and the cover image for the identified profile.

Wherein, the update event of the cover image includes at least one of the following:

In the case of updating files;

In the case of obtaining the update instruction of the cover image;

In the case of periodically updating the cover image.

It can be seen that in the embodiment of the present disclosure, when it is determined to update the file, or when it is determined that the cover image needs to be updated, by performing steps 101 to 104, the cover image can be refreshed; in the case of periodic update of the cover image, By performing steps 101 to 104, the continuous update of the cover image can be automatically realized, and since there is no need to manually correct the cover image, the operation and maintenance cost can be reduced; in some embodiments, when the file is updated, the image in the file is Recently acquired images, that is, the embodiment of the present disclosure can select a cover image based on a recently acquired image, so that the selected cover image can reflect the recent state of the target object, thereby facilitating subsequent feature retrieval for the target object.

In some embodiments of the present disclosure, referring to FIG. 7 , the process of acquiring the image feature set of the target object may include:

Step 701: Determine the duration between the acquisition time and the current time of each image in the file corresponding to the target object.

Step 702: In the archive, filter out images whose duration exceeds the life cycle of the image to obtain an updated archive.

In practical applications, the spark computing engine can be used to filter out the images in the archives whose time between the acquisition time and the current time exceeds the life cycle of the image, so as to obtain an updated archive.

Step 703: Perform feature extraction on the updated archive to obtain a set of image features of the target object in the archive.

In some embodiments, since the images in the archive consume a lot of disk resources, the image survival time IMAGE_URL_TTL can be set for the image, similarly, a feature survival time FEATURE_TTL can also be set for the features in the bottom library; for example, The smaller value of the image survival time IMAGE_URL_TTL and the feature survival time FEATURE_TTL can be used as the survival period of the image.

In an implementation manner, the survival period may be one week, 15 days, one month, and so on.

Understandably, it is not possible to select an image whose duration between the acquisition time and the current time in the archive exceeds the life cycle of the image, otherwise the selected cover image will expire, or the features in the base library will expire; in the embodiment of the present disclosure, the Filtering out images whose duration between the collection time and the current time exceeds the life cycle of the image in the archive can improve the reliability of the process of selecting a cover image; in some embodiments, the embodiments of the present disclosure can select representative images based on the most recently collected images image, so that the selected representative image can reflect the latest state of the target object.

In some embodiments of the present disclosure, before updating the cover image, referring to FIG. 8 , the above file processing method may further include:

Step 801 : In response to adding a new image of a target object to the dossier, determine the number of images in the dossier.

Step 802: In response to the fact that the number of images in the archive exceeds the preset number, sequentially filter out the images in the archive according to the order of collection time in the archive from early to late until the number of images in the archive is less than or equal to the preset number until.

Here, in response to the fact that the number of images in the archive does not exceed a preset number, the images in the archive may remain unchanged.

In some embodiments of the present disclosure, after the cover image of the archive is determined, the cover image may be displayed, and the cover image may also be analyzed or feature retrieved. Exemplarily, after acquiring the feature retrieval request of any object, according to the feature retrieval request, the feature retrieval can be performed on the cover image of the files corresponding to different objects.

It can be seen that since the cover image has a good image representation, the accuracy of the feature retrieval can be improved by retrieving the features of the cover image. For feature retrieval, the embodiments of the present disclosure perform feature retrieval based on cover images, which can effectively reduce the number of images to be retrieved and improve retrieval efficiency. In addition, integrating retrieval results of multiple cover images can also effectively improve the recall rate of retrieval.

In some scenarios, the image data captured by the camera will be extracted and clustered to form a file corresponding to the target object. In related technologies, when the user checks the cover image by querying the file, the image may not be accessible or the image may be difficult to identify , it is necessary to further click on the file details to find an image with better imaging quality for comparison, in order to confirm whether it is the file corresponding to the target object. However, after adopting the technical solutions of the embodiments of the present disclosure, the output cover image is a typical snapshot image in file snapshots, and the files of non-target objects can be filtered by using the cover image.

On the basis of the file processing methods provided in the foregoing embodiments, the embodiments of the present disclosure also provide a file processing device.

Fig. 9 is a schematic structural diagram of a file processing device according to an embodiment of the present disclosure. As shown in Fig. 9, the device may include: an acquisition part 900, a first processing part 901, a second processing part 902 and a third processing part 903, wherein,

The acquiring part 900 is configured to acquire the image feature set of the target object;

The first processing part 901 is configured to determine a similarity set between each image feature in the image feature set and other image features in the image feature set;

The second processing part 902 is configured to determine representative features of the image feature set according to the similarity set;

The third processing part 903 is configured to determine the image corresponding to the representative feature as the cover image of the file corresponding to the target object.

In some embodiments, the second processing part 902 is configured to determine representative features of the image feature set according to the similarity set, including:

determining the local density of each image feature according to the similarity set;

Determine the density point interval of each image feature according to the similarity set and the local density of each image feature;

A representative feature of the image feature set is determined according to the local density and the density point interval of each image feature.

In some embodiments, the second processing part 902 is configured to determine the local density of each image feature according to the similarity set, including:

In the set of image features, determine at least one image feature whose similarity with each image feature is greater than a threshold; use the sum of the similarities between the at least one image feature and each image feature as the The local density of each image feature.

In some embodiments, the second processing part 902 is configured to determine the density point interval of each image feature according to the similarity set and the local density of each image feature, including:

In the case that the first image feature in the image feature set is not the image feature with the largest local density in the image feature set, select an image feature with a local density greater than the local density of the first image feature in the image feature set Image feature, taking the minimum value of the similarity between the selected image feature and the first image feature as the density point interval of the first image feature; the first image feature is any of the image feature sets An image feature; in the case of obtaining the density point interval of each image feature in the image feature set except the image feature with the local maximum density, in the image feature set except the image feature with the local maximum density In the density point interval of each image feature other than the feature, the minimum value of the density point interval is used as the density point interval of the image feature with the highest local density in the image feature set.

In some embodiments, the second processing part 902 is configured to determine representative features of the image feature set according to the local density and the density point interval of each image feature, including:

According to the local density and the density point interval of each image feature, a representative score of each image feature is obtained, and the representative score of each image feature is positively correlated with the local density, Negatively correlated with the density point interval;

In the image feature set, at least one feature whose representative score is greater than or equal to a representative score threshold is selected as the representative feature.

In some embodiments, the second processing part 902 is configured to obtain the representative score of each image feature according to the local density and the density point interval of each image feature, including:

The ratio of the local density of each image feature to the density point interval is used as the representative score of each image feature.

In some embodiments, the third processing part 903 is further configured to perform the following steps in response to an update event of the cover image after determining the cover image of the archive for the first time, to implement the cover image of the archive Image updating: acquiring the image feature set of the target object, determining representative features of the image feature set, and determining the cover image of the file; wherein:

The update event of the cover image includes at least one of the following:

in the case of updating said file;

In the case of acquiring the update instruction of the cover image;

In the case of periodically updating the cover image.

In some embodiments, the acquisition part 900 is configured to acquire the image feature set of the target object, including:

Determining the acquisition time of each image in the file corresponding to the target object and the duration of the current time;

In the archive, the images whose duration exceeds the life cycle of the image are filtered out to obtain an updated archive;

Feature extraction is performed on the updated archive to obtain an image feature set of the target object.

In some embodiments, the acquiring portion 900 is further configured to determine the number of images in the archive in response to adding a new image of the target object to the archive before updating the cover image; In response to the fact that the number of images in the archive exceeds a preset number, sequentially filter out the images in the archive according to the order of collection time in the archive from early to late until the number of images in the archive is less than or equal to the preset amount.

The acquisition part 900, the first processing part 901, the second processing part 902 and the third processing part 903 mentioned above can be realized based on the processor of the electronic device.

In addition, each functional module in this embodiment may be integrated into one processing unit, each unit may exist separately physically, or two or more units may be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software function modules.

If the integrated unit is implemented in the form of a software function module and is not sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this embodiment is essentially or It is said that the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product, the computer software product is stored in a storage medium, and includes several instructions to make a computer device (which can It is a personal computer, a server, or a network device, etc.) or a processor (processor) that executes all or part of the steps of the method described in this embodiment. The aforementioned storage medium includes: U disk, mobile hard disk, read only memory (Read Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disk or optical disk and other various media that can store program codes.

Specifically, the computer program instructions corresponding to a file processing method in this embodiment can be stored on a storage medium such as an optical disc, a hard disk, or a USB flash drive. When the computer program instructions corresponding to a file processing method in the storage medium When read or executed by an electronic device, any one of the file processing methods of the foregoing embodiments is realized.

Based on the same technical idea of the foregoing embodiments, see FIG. 10 , which shows an electronic device 1000 provided by an embodiment of the present disclosure, which may include: a memory 1001, a processor 1002, and an A computer program running on ; where,

memory 1001 for storing computer programs and data;

The processor 1002 is configured to execute the computer program stored in the memory, so as to realize any file processing method of the foregoing embodiments.

In practical applications, the above-mentioned memory 1001 can be a volatile memory (volatile memory), such as RAM; or a non-volatile memory (non-volatile memory), such as ROM, flash memory (flash memory), hard disk (Hard Disk Drive, HDD) or solid-state drive (Solid-State Drive, SSD); or a combination of the above-mentioned types of memory, and provide instructions and data to the processor 1002.

The aforementioned processor 1002 may be at least one of ASIC, DSP, DSPD, PLD, FPGA, CPU, controller, microcontroller, and microprocessor.

In some embodiments, the functions or parts included in the device provided by the embodiments of the present disclosure can be used to execute the methods described in the method embodiments above, and its specific implementation can refer to the description of the method embodiments above. For brevity, here No longer.

The above descriptions of the various embodiments tend to emphasize the differences between the various embodiments, and the same or similar points can be referred to each other. For the sake of brevity, no further details are given here.

The methods disclosed in the various method embodiments provided in the present disclosure can be combined arbitrarily without conflict to obtain new method embodiments.

The features disclosed in the various product embodiments provided in the present disclosure can be combined arbitrarily without conflict to obtain new product embodiments.

The features disclosed in each method or device embodiment provided in the present disclosure can be combined arbitrarily without conflict to obtain a new method embodiment or device embodiment.

Through the description of the above embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus a necessary general-purpose hardware platform, and of course also by hardware, but in many cases the former is better implementation. Based on such an understanding, the essence of the technical solution of the present invention or the part that contributes to the prior art can be embodied in the form of software products, and the computer software products are stored in a storage medium (such as ROM/RAM, disk, CD) contains several instructions to enable a terminal (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in various embodiments of the present disclosure.

The embodiments of the present disclosure have been described above in conjunction with the accompanying drawings, but the present disclosure is not limited to the above-mentioned specific implementations. The above-mentioned specific implementations are only illustrative and not restrictive. Under the inspiration of the present disclosure, without departing from the purpose of the present disclosure and the protection scope of the claims, many forms can also be made, and these all belong to the protection of the present disclosure.

Claims

A file processing method applied to electronic equipment, the method comprising:

Obtain a set of image features of the target object;

determining a set of similarities between each image feature in the set of image features and other image features in the set of image features;

Determine representative features of the image feature set according to the similarity set;

The image corresponding to the representative feature is determined as the cover image of the file corresponding to the target object.
The method according to claim 1, wherein said determining representative features of said image feature set according to said similarity set comprises:

determining the local density of each image feature according to the similarity set;

Determine the density point interval of each image feature according to the similarity set and the local density of each image feature;

A representative feature of the image feature set is determined according to the local density and the density point interval of each image feature.
The method according to claim 2, wherein said determining the local density of each image feature according to said similarity set comprises:

In the set of image features, determining at least one image feature whose similarity with each image feature is greater than a threshold;

The sum of similarities between the at least one image feature and each image feature is used as the local density of each image feature.
The method according to claim 2 or 3, wherein said determining the density point interval of each image feature according to the similarity set and the local density of each image feature comprises:

In the case that the first image feature in the image feature set is not the image feature with the largest local density in the image feature set, select an image feature with a local density greater than the local density of the first image feature in the image feature set Image feature, taking the minimum value of the similarity between the selected image feature and the first image feature as the density point interval of the first image feature; the first image feature is any of the image feature sets an image feature;

In the case of obtaining the density point spacing of each image feature in the image feature set except for the image feature with the highest local density, each image feature in the set of image features except the image feature with the highest local density In the density point interval of the image feature, the minimum value of the density point interval is used as the density point interval of the image feature with the highest local density in the image feature set.
The method according to any one of claims 2 to 4, wherein said determining representative features of said image feature set according to said local density and said density point interval of said each image feature comprises:

According to the local density and the density point interval of each image feature, a representative score of each image feature is obtained, and the representative score of each image feature is positively correlated with the local density, Negatively correlated with the density point interval;

In the image feature set, at least one feature whose representative score is greater than or equal to a representative score threshold is selected as the representative feature.
The method according to claim 5, wherein the representative score of each image feature is obtained according to the local density and the density point interval of each image feature, comprising:

The ratio of the local density of each image feature to the density point interval is used as the representative score of each image feature.
The method according to any one of claims 1 to 6, wherein, after first determining the cover image of the profile, the method further comprises:

In response to the update event of the cover image, the update of the cover image of the archive is realized by performing the following steps:

acquiring a set of image features of the target object, determining a representative feature of the set of image features, and determining a cover image of the profile; wherein:

The update event of the cover image includes at least one of the following:

in the case of updating said file;

In the case of acquiring the update instruction of the cover image;

In the case of periodically updating the cover image.
The method according to any one of claims 1 to 7, wherein said acquiring the image feature set of the target object comprises:

Determining the acquisition time of each image in the file corresponding to the target object and the duration of the current time;

In the archive, the images whose duration exceeds the life cycle of the image are filtered out to obtain an updated archive;

Feature extraction is performed on the updated archive to obtain an image feature set of the target object.
The method according to any one of claims 1 to 8, wherein, before updating the cover image, the method further comprises:

determining a number of images in the dossier in response to adding a new image of the target object to the dossier;

In response to the fact that the number of images in the archive exceeds a preset number, sequentially filter out the images in the archive according to the order of collection time in the archive from early to late until the number of images in the archive is less than or equal to the preset amount.
An archive processing device, the device comprising an acquisition part, a first processing part, a second processing part and a third processing part, wherein,

The acquisition part is used to acquire the image feature set of the target object;

A first processing part, configured to determine a similarity set between each image feature in the image feature set and other image features in the image feature set;

The second processing part is used to determine representative features of the image feature set according to the similarity set;

The third processing part is configured to determine the image corresponding to the representative feature as the cover image of the file corresponding to the target object.
The device according to claim 10, wherein the second processing part is configured to determine representative features of the image feature set according to the similarity set, comprising:

determining the local density of each image feature according to the similarity set;

Determine the density point interval of each image feature according to the similarity set and the local density of each image feature;

A representative feature of the image feature set is determined according to the local density and the density point interval of each image feature.
The device according to claim 11, wherein the second processing part is configured to determine the local density of each image feature according to the similarity set, comprising:

In the set of image features, determining at least one image feature whose similarity with each image feature is greater than a threshold;

The sum of similarities between the at least one image feature and each image feature is used as the local density of each image feature.
The device according to claim 11 or 12, wherein the second processing part is configured to determine the density point interval of each image feature according to the similarity set and the local density of each image feature, comprising :

In the case that the first image feature in the image feature set is not the image feature with the largest local density in the image feature set, select an image feature with a local density greater than the local density of the first image feature in the image feature set Image feature, taking the minimum value of the similarity between the selected image feature and the first image feature as the density point interval of the first image feature; the first image feature is any of the image feature sets an image feature;

In the case of obtaining the density point spacing of each image feature in the image feature set except for the image feature with the highest local density, each image feature in the set of image features except the image feature with the highest local density In the density point interval of the image feature, the minimum value of the density point interval is used as the density point interval of the image feature with the highest local density in the image feature set.
The device according to any one of claims 11 to 13, wherein the second processing part is configured to determine the image feature set according to the local density and the density point interval of each image feature representative features, including:

According to the local density and the density point interval of each image feature, a representative score of each image feature is obtained, and the representative score of each image feature is positively correlated with the local density, Negatively correlated with the density point interval;

In the image feature set, at least one feature whose representative score is greater than or equal to a representative score threshold is selected as the representative feature.
The device according to claim 14, wherein the second processing part is configured to obtain the representativeness of each image feature according to the local density and the density point spacing of each image feature. scores, including:

The ratio of the local density of each image feature to the density point interval is used as the representative score of each image feature.
The device according to any one of claims 10 to 15, wherein the third processing part is further configured to respond to an update event of the cover image by executing the following after the cover image of the archive is determined for the first time Step, realizing the update of the cover image of the archive: obtaining the image feature set of the target object, determining the representative feature of the image feature set, and determining the cover image of the archive;

The update event of the cover image includes at least one of the following:

in the case of updating said file;

In the case of acquiring the update instruction of the cover image;

In the case of periodically updating the cover image.
The device according to any one of claims 10 to 16, wherein the acquiring part is configured to acquire a set of image features of the target object, comprising:

Determining the acquisition time of each image in the file corresponding to the target object and the duration of the current time;

In the archive, the images whose duration exceeds the life cycle of the image are filtered out to obtain an updated archive;

Feature extraction is performed on the updated archive to obtain an image feature set of the target object.
The apparatus according to any one of claims 10 to 17, wherein the acquiring part is further configured to respond to adding a new image of the target object to the archive before updating the cover image, determining the number of images in the archive; in response to the fact that the number of images in the archive exceeds a preset number, sequentially filtering out the images in the archive according to the sequence of collection time in the archive from early to late, until the number of images in the file is less than or equal to the preset number.
An electronic device comprising a processor and a memory for storing a computer program capable of running on the processor; wherein,

The processor is used to run the computer program to execute the file processing method described in any one of claims 1-9.
A computer storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the file processing method described in any one of claims 1 to 9 is realized.
A computer program, comprising computer-readable codes, when the computer-readable codes are run in an electronic device, a processor in the electronic device executes the archive processing described in any one of claims 1 to 9 method.