WO2019011275A1 - Video search method and apparatus, cloud storage system and computer device - Google Patents

Abstract

Provided are a video search method and apparatus, a cloud storage system and a computer device. The method comprises: a management node receiving a video search request, including feature information about a surveillance video to be searched, sent by a client, wherein the feature information comprises at least one of the following: a reference picture, a spatial range, a time range and attribute information; the management node determining a target storage node storing the surveillance video to be searched, and sending the video search request to each target storage node; each target storage node receiving the video search request, searching, according to a locally stored surveillance video, stored surveillance video index information and the feature information, for target surveillance video index information with a matching degree between same and the feature information being greater than a first pre-set threshold value, and sending the target surveillance video index information to the management node. The embodiments of the present application can realize the search of a surveillance video in a cloud storage system.

Description

Video search method, device, cloud storage system and computer equipment

This application claims priority to the Chinese Patent Application entitled "A Video Search Method, Apparatus, Cloud Storage System, and Computer Equipment" filed on July 12, 2017, with the application number of 201710565416.1, the entire contents of which are hereby incorporated by reference. The citations are incorporated herein by reference.

Technical field

The present application relates to the field of video surveillance technologies, and in particular, to a video search method, apparatus, cloud storage system, and computer device.

Background technique

With the continuous development of video surveillance technology, the construction of safe cities, intelligent communities, and intelligent transportation is becoming more and more popular, and video surveillance applications are becoming more and more widespread. This will result in an explosive increase in the geometrical index of the surveillance video.

In order to meet the storage of massive surveillance video, cloud storage systems have emerged. Specifically, the cloud storage system may include a management node and multiple storage nodes, and the monitoring video collected by the front-end monitoring device needs to be written into the cloud storage system through the access node. In the process of monitoring video storage, as shown in Figure 1, the data flow direction is 1.1-1.2-1.3-1.4, that is, when the access node specifically accesses the monitoring video of any front-end monitoring device, it can first manage to the access node. The node applies to allocate a storage node to the front-end monitoring device, so that the front-end monitoring device can send the collected monitoring video to the storage node for storage.

The management node stores the index information of the monitoring video stored in each storage node, where the index information includes at least the rough location information corresponding to the monitoring video stored in each storage node, such as the province, city, and region information; The storage node also stores index information of the monitoring video stored by the storage node, where the index information includes at least time information and detailed location information corresponding to each monitoring video, such as the name of the street to which it belongs.

However, just providing the ability to store a large amount of surveillance video is not enough for the actual application of the user. How to intelligently and quickly find the monitoring video of interest and interest of the user in the massive monitoring video stored in the cloud storage system, such as a monitoring video located at a certain location within a certain period of time, or a specific target Monitoring video has become an urgent problem to be solved.

Summary of the invention

The purpose of the embodiments of the present application is to provide a video search method, device, cloud storage system, and computer device to implement monitoring of monitoring video in a cloud storage system. The specific technical solutions are as follows:

In a first aspect, an embodiment of the present application provides a video search method, where the method includes:

The management node receives a video search request that is sent by the client and includes feature information of the to-be-searched surveillance video, where the feature information includes at least one of the following: a reference picture, a spatial range, a time range, and attribute information;

The management node determines a target storage node that stores the to-be-searched surveillance video, and sends the video search request to each of the target storage nodes;

Each of the target storage nodes receives the video search request, and according to the locally stored monitoring video, the stored monitoring video index information, and the feature information, the matching degree between the search and the feature information is greater than the first preset. The target of the threshold monitors the video index information, and sends the target monitoring video index information to the management node.

Optionally, when the feature information includes: a reference picture, a spatial range, a time range, and attribute information, each of the target storage nodes is configured according to locally stored monitoring video, stored monitoring video index information, and And the step of searching for the target monitoring video index information that the matching degree with the feature information is greater than the first preset threshold includes:

Performing feature extraction on the reference picture by any target storage node to obtain picture feature information corresponding to the reference picture;

The target storage node searches for the candidate monitoring video index information that matches the spatial range and the time range in the locally stored monitoring video index information;

The target storage node sequentially reads the monitoring video corresponding to each candidate monitoring video index information in the candidate monitoring video index information, and extracts video feature information of the monitoring video by using a pre-configured feature algorithm, according to the video of the monitoring video. The feature information, and the picture feature information and the attribute information, calculate a matching degree between the candidate monitoring video index information and the feature information, and determine whether the matching degree is greater than a first preset threshold, and if so, The candidate monitor video index information is used as the target monitoring video index information.

Optionally, the target storage node sequentially reads the monitoring video corresponding to each candidate monitoring video index information in the candidate monitoring video index information, and extracts video feature information of the monitoring video by using a pre-configured feature algorithm. Calculating, according to the video feature information of the monitoring video, the image feature information and the attribute information, a matching degree between the candidate monitoring video index information and the feature information, and determining whether the matching degree is greater than a first preset threshold If yes, the step of using the candidate monitoring video index information as the target monitoring video index information includes:

The target storage node constructs an initial index information set including each candidate monitoring video index information, and constructs an empty result index information set;

Determining whether the initial index information set is empty;

If not, the monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the initial index information set, and the video feature information of the monitoring video is extracted by using a pre-configured feature algorithm, and according to the monitoring video The video feature information, and the image feature information and the attribute information, calculate a matching degree between the candidate monitoring video index information and the feature information, and determine whether the matching degree is greater than a first preset threshold, and if yes, Adding the candidate monitoring video index information to the result index information set, deleting the candidate monitoring video index information from the initial index information set, and returning to perform the step of determining whether the initial index information set is empty;

If yes, the index information included in the result index information set is used as the target monitoring video index information.

Optionally, each of the storage nodes is pre-configured with a plurality of feature algorithms, and each feature algorithm has a corresponding priority. The step of extracting the video feature information of the monitoring video by using the pre-configured feature algorithm includes:

Extracting video feature information of the surveillance video by using a feature algorithm with the highest priority;

The step of using the index information included in the result index information set as the target monitoring video index information includes:

Determining whether the result index information set is empty; when the result index information set is empty, the target storage node again constructs an initial index information set including each candidate monitoring video index information, and determines the newly constructed initial index information set. Whether it is empty; when the result index information set is not empty, the index information included in the result index information set is used as the target monitoring video index information;

When the newly constructed initial index information set is not empty, the monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the newly constructed initial index information set, and the next priority feature algorithm is adopted. Extracting video feature information of the monitoring video, and calculating a matching degree between the candidate monitoring video index information and the feature information according to the video feature information of the monitoring video, the image feature information, and the attribute information, and determining If the matching degree is greater than the first preset threshold, if yes, the candidate monitoring video index information is added to the result index information set, and the candidate monitoring video index information is deleted from the newly constructed initial index information set. And returning to the step of performing the determining whether the newly constructed initial index information set is empty;

When the newly constructed initial index information set is empty, a step of performing the determination as to whether the result index information set is empty is returned.

Optionally, after the sending the target monitoring video index information to the management node, the method further includes:

The management node receives target monitoring video index information sent by each target storage node, and sends the target monitoring video index information to the client.

Optionally, after the sending the target monitoring video index information to the client, the method further includes:

The management node receives the confirmation information sent by the client, and adjusts a first preset threshold in each target storage node according to the confirmation information.

Optionally, the step of adjusting the first preset threshold in each target storage node according to the confirmation information includes:

When the accuracy rate included in the confirmation information is greater than a second preset threshold, decreasing the first preset threshold;

When the accuracy rate included in the confirmation information is less than or equal to the second threshold, the first preset threshold is increased.

Optionally, the step of determining, by the management node, that the target storage node that is to be searched for the monitored video is stored includes:

The management node determines, according to the spatial range and the index information of the monitoring video stored in each locally stored storage node, the target storage node that stores the to-be-searched surveillance video.

Optionally, the step of sending the video search request to each of the target storage nodes includes:

The management node constructs a list of target storage nodes including identifiers of the target storage nodes;

Determining whether the target storage node list is empty; when the target storage node list is not empty, sending the video search request to a target storage node in the target storage node list, and determining whether the transmission is successful;

When the sending is successful, removing the target storage node identifier from the target storage node list, and returning to perform the step of determining whether the target storage node list is empty;

When the transmission fails, a step of performing the determination of whether the storage node list is empty is returned.

Optionally, before the management node receives the video search request that is sent by the client and includes the feature information of the to-be-searched video, the method further includes:

The client receives the reference picture uploaded by a user;

The client receives the spatial range, the time range, and the attribute information input by a user;

The client integrates the reference picture, the spatial range, the time range, and the attribute information to generate the video search request, and sends the video search request to the management node.

In a second aspect, the embodiment of the present application provides a video search method, which is applied to a target storage node in a cloud storage system, where the cloud storage system further includes a management node, and the method includes:

Receiving, by the management node, a video search request that includes feature information of a to-be-searched surveillance video, where the video search request is that the management node determines that the target storage node stores a to-be-searched surveillance video, and stores the target information The feature information sent by the node includes at least one of the following: a reference picture, a spatial range, a time range, and attribute information;

Searching, according to the locally stored monitoring video, the stored monitoring video index information, and the feature information, target monitoring video index information that is greater than a first preset threshold between the feature information and the target The monitoring video index information is sent to the management node.

Optionally, when the feature information includes: a reference picture, a spatial range, a time range, and attribute information, the searching and storing according to the locally stored monitoring video, the stored monitoring video index information, and the feature information, The step of matching the feature information with the target monitoring video index information that is greater than the first preset threshold includes:

Performing feature extraction on the reference picture to obtain picture feature information corresponding to the reference picture;

Finding candidate monitoring video index information that matches the spatial range and time range in the locally stored monitoring video index information;

In the candidate monitoring video index information, the monitoring video corresponding to each candidate monitoring video index information is sequentially read, and the video feature information of the monitoring video is extracted by using a pre-configured feature algorithm, according to the video feature information of the monitoring video, and Determining, by the picture feature information and the attribute information, a matching degree between the candidate monitoring video index information and the feature information, and determining whether the matching degree is greater than a first preset threshold, and if yes, selecting the candidate monitoring video The index information is used as the target monitoring video index information.

Optionally, in the candidate monitoring video index information, the monitoring video corresponding to each candidate monitoring video index information is sequentially read, and the video feature information of the monitoring video is extracted by using a pre-configured feature algorithm, according to the monitoring video. The video feature information, and the image feature information and the attribute information, calculate a matching degree between the candidate monitoring video index information and the feature information, and determine whether the matching degree is greater than a first preset threshold, and if so, The step of using the candidate monitoring video index information as the target monitoring video index information includes:

Constructing an initial index information set including each candidate monitoring video index information, and constructing an empty result index information set;

Determining whether the initial index information set is empty;

If not, the monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the initial index information set, and the video feature information of the monitoring video is extracted by using a pre-configured feature algorithm, and according to the monitoring video The video feature information, and the image feature information and the attribute information, calculate a matching degree between the candidate monitoring video index information and the feature information, and determine whether the matching degree is greater than a first preset threshold, and if yes, Adding the candidate monitoring video index information to the result index information set, deleting the candidate monitoring video index information from the initial index information set, and returning to perform the step of determining whether the initial index information set is empty;

If yes, the index information included in the result index information set is used as the target monitoring video index information.

Optionally, the target storage node is pre-configured with a plurality of feature algorithms, and each feature algorithm has a corresponding priority. The step of extracting the video feature information of the monitoring video by using the pre-configured feature algorithm includes:

Extracting video feature information of the surveillance video by using a feature algorithm with the highest priority;

The step of using the index information included in the result index information set as the target monitoring video index information includes:

Determining whether the result index information set is empty; when the result index information set is empty, the target storage node again constructs an initial index information set including each candidate monitoring video index information, and determines the newly constructed initial index information set. Whether it is empty;

When the newly constructed initial index information set is not empty, the monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the newly constructed initial index information set, and the next priority feature algorithm is adopted. Extracting video feature information of the monitoring video, and calculating a matching degree between the candidate monitoring video index information and the feature information according to the video feature information of the monitoring video, the image feature information, and the attribute information, and determining If the matching degree is greater than the first preset threshold, if yes, the candidate monitoring video index information is added to the result index information set, and the candidate monitoring video index information is deleted from the newly constructed initial index information set. And returning to the step of performing the determining whether the newly constructed initial index information set is empty;

When the newly constructed initial index information set is empty, returning to perform the step of determining whether the result index information set is empty;

When the result index information set is not empty, the index information included in the result index information set is used as the target monitoring video index information.

In a third aspect, the embodiment of the present application provides a cloud storage system, where the cloud storage system includes: a management node and multiple storage nodes.

The management node is configured to receive, by the client, a video search request that includes feature information of the to-be-searched surveillance video, where the feature information includes at least one of: a reference picture, a spatial range, a time range, and attribute information; Determining a target storage node storing the to-be-searched surveillance video, and transmitting the video search request to each of the target storage nodes;

Each of the target storage nodes is configured to receive the video search request, and according to the locally stored monitoring video, the stored monitoring video index information, and the feature information, the matching degree between the search and the feature information is greater than A target of a preset threshold monitors video index information, and sends the target monitoring video index information to the management node.

Optionally, when the feature information includes: a reference picture, a spatial range, a time range, and attribute information,

Any target storage node is specifically configured to perform feature extraction on the reference picture to obtain picture feature information corresponding to the reference picture; and search for candidate matching the spatial range and time range in the locally stored monitoring video index information Monitoring video index information; in the candidate monitoring video index information, sequentially reading the monitoring video corresponding to each candidate monitoring video index information, and extracting video feature information of the monitoring video by using a pre-configured feature algorithm, according to the monitoring video The video feature information, and the image feature information and the attribute information, calculate a matching degree between the candidate monitoring video index information and the feature information, and determine whether the matching degree is greater than a first preset threshold, and if yes, The candidate monitoring video index information is used as the target monitoring video index information.

Optionally, any target storage node is specifically configured to construct an initial index information set including each candidate monitoring video index information, and construct an empty result index information set; determine whether the initial index information set is empty; if not, The monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the initial index information set, and the video feature information of the monitoring video is extracted by using a pre-configured feature algorithm, and according to the video feature information of the monitoring video. And the picture feature information and the attribute information, calculating a matching degree between the candidate monitoring video index information and the feature information, and determining whether the matching degree is greater than a first preset threshold, and if yes, selecting the candidate The monitoring video index information is added to the result index information set, the candidate monitoring video index information is deleted from the initial index information set, and a step of determining whether the initial index information set is empty is performed; if yes, Taking the index information included in the result index information set as a The target monitors the video index information.

Optionally, multiple feature algorithms are pre-configured in each storage node, and each feature algorithm has a corresponding priority.

Any target storage node is specifically configured to extract video feature information of the monitoring video by using a feature algorithm with the highest priority; and when the index information included in the result index information set is used as the target monitoring video index information, Whether the result index information set is empty; when the result index information set is empty, the target storage node again constructs an initial index information set including each candidate monitoring video index information, and determines whether the newly constructed initial index information set is air;

When the newly constructed initial index information set is not empty, the monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the newly constructed initial index information set, and the next priority feature algorithm is adopted. Extracting video feature information of the monitoring video, and calculating a matching degree between the candidate monitoring video index information and the feature information according to the video feature information of the monitoring video, the image feature information, and the attribute information, and determining If the matching degree is greater than the first preset threshold, if yes, the candidate monitoring video index information is added to the result index information set, and the candidate monitoring video index information is deleted from the newly constructed initial index information set. And returning to the step of performing the determining whether the newly constructed initial index information set is empty;

When the newly constructed initial index information set is empty, returning to perform the step of determining whether the result index information set is empty;

When the result index information set is not empty, the index information included in the result index information set is used as the target monitoring video index information.

Optionally, the management node is further configured to receive target monitoring video index information sent by each target storage node, and send the target monitoring video index information to the client.

Optionally, the management node is further configured to receive the acknowledgement information sent by the client, and adjust the first preset threshold in each target storage node according to the acknowledgement information.

Optionally, the management node is specifically configured to: when the accuracy rate included in the confirmation information is greater than a second preset threshold, reduce the first preset threshold; when the accuracy included in the confirmation information When less than or equal to the second threshold, the first preset threshold is increased.

Optionally, the management node is configured to determine, according to the spatial range, and the index information of the monitoring video stored in each locally stored storage node, the target storage node that stores the to-be-searched surveillance video.

Optionally, the management node is specifically configured to construct a target storage node list that includes each target storage node identifier; determine whether the target storage node list is empty; and when the target storage node list is not empty, Sending a video search request to a target storage node in the target storage node list, and determining whether the transmission is successful; when the transmission is successful, removing the target storage node identifier from the target storage node list, and returning to perform the Determining whether the target storage node list is empty; when the transmission fails, returning to the step of performing the determining whether the storage node list is empty.

In a fourth aspect, the embodiment of the present application provides a video search device, which is applied to a target storage node in a cloud storage system, where the cloud storage system further includes a management node, and the device includes:

a receiving module, configured to receive, by the management node, a video search request that includes feature information of a to-be-searched surveillance video, where the video search request is that the management node determines that the target storage node stores a to-be-searched surveillance video, Sending to the target storage node, the feature information includes at least one of the following: a reference picture, a spatial range, a time range, and attribute information;

a search module, configured to search, according to the locally stored monitoring video, the stored monitoring video index information, and the feature information, a target monitoring video index information that is greater than a first preset threshold between the feature information and the feature information, And transmitting the target monitoring video index information to the management node.

Optionally, when the feature information includes: a reference picture, a spatial range, a time range, and attribute information, the searching module includes:

a feature extraction sub-module, configured to perform feature extraction on the reference picture to obtain picture feature information corresponding to the reference picture;

a search submodule, configured to search, in the locally stored monitoring video index information, candidate monitoring video index information that matches the spatial range and time range;

The search sub-module is configured to sequentially read the monitoring video corresponding to each candidate monitoring video index information in the candidate monitoring video index information, and extract the video feature information of the monitoring video by using a pre-configured feature algorithm, according to the monitoring video. The video feature information, and the image feature information and the attribute information, calculate a matching degree between the candidate monitoring video index information and the feature information, and determine whether the matching degree is greater than a first preset threshold, and if so, The candidate monitoring video index information is used as the target monitoring video index information.

Optionally, the search submodule includes:

Constructing a subunit for constructing an initial index information set including each candidate monitoring video index information, and constructing an empty result index information set;

a determining subunit, configured to determine whether the initial index information set is empty;

a feature extraction subunit, configured to sequentially read, in the initial index information set, a monitoring video corresponding to a piece of candidate monitoring video index information in the initial index information set, using pre-configured features, when the determining subunit determines that the result is negative The algorithm extracts video feature information of the surveillance video;

a search subunit, configured to calculate a matching degree between the candidate monitoring video index information and the feature information according to the video feature information of the monitoring video, and the image feature information and the attribute information, and determine whether the matching degree is If the value is greater than the first preset threshold, if yes, the candidate monitoring video index information is added to the result index information set, the candidate monitoring video index information is deleted from the initial index information set, and the determining subunit is triggered. ;

Determining a subunit, configured to use index information included in the result index information set as the target monitoring video index information when the judgment subunit determines that the result is YES.

Optionally, the feature extraction sub-unit is specifically configured to extract video feature information of the surveillance video by using a feature algorithm with the highest priority;

The determining subunit is specifically configured to:

Determining whether the result index information set is empty; when the result index information set is empty, the target storage node again constructs an initial index information set including each candidate monitoring video index information, and determines the newly constructed initial index information set. Whether it is empty;

When the newly constructed initial index information set is not empty, the monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the newly constructed initial index information set, and the next priority feature algorithm is adopted. Extracting video feature information of the monitoring video, and calculating a matching degree between the candidate monitoring video index information and the feature information according to the video feature information of the monitoring video, the image feature information, and the attribute information, and determining If the matching degree is greater than the first preset threshold, if yes, the candidate monitoring video index information is added to the result index information set, and the candidate monitoring video index information is deleted from the newly constructed initial index information set. And returning to the step of performing the determining whether the newly constructed initial index information set is empty;

When the newly constructed initial index information set is empty, returning to perform the step of determining whether the result index information set is empty;

When the result index information set is not empty, the index information included in the result index information set is used as the target monitoring video index information.

In a fifth aspect, an embodiment of the present application provides a computer device, including a processor, a communication interface, a memory, and a communication bus, where the processor, the communication interface, and the memory complete each other through the communication bus. Communication;

The memory is configured to store a computer program;

The processor is configured to implement the method steps described in the second aspect when the program stored on the memory is executed.

In a sixth aspect, the present application provides a computer readable storage medium, wherein the computer readable storage medium is for storing executable program code for performing the second aspect of the present application at runtime Method steps described.

In a seventh aspect, the application provides an application, wherein the application is operative to perform the method steps of the second aspect of the present application at runtime.

In the embodiment of the present application, when the management node receives the video search request, it may first determine the target storage node storing the video to be searched, thereby transmitting the video search request to each target storage node, and each target storage node receives the video search request. After that, in the monitoring video stored by itself, the target monitoring video index information matching the video search request may be searched, and the searched target monitoring video index information is sent to the management node, so that the management node receives the The video search request matches all the target monitoring video index information, and realizes the search of the monitoring video in the cloud storage system. Moreover, each storage node separately searches for the target monitoring video index information stored by itself and matches the video search request, thereby ensuring the efficiency of monitoring the video search. Of course, the embodiments of the present application do not necessarily achieve all of the above advantages.

DRAWINGS

In order to more clearly illustrate the embodiments of the present application and the technical solutions of the prior art, the following description of the embodiments and the drawings used in the prior art will be briefly introduced. Obviously, the drawings in the following description are only Some embodiments of the application may also be used to obtain other figures from those of ordinary skill in the art without departing from the scope of the invention.

FIG. 1 is a schematic diagram of a process of storing a monitoring video to a cloud storage system;

2 is a system architecture diagram of a video search method to which an embodiment of the present application is applied;

3 is a flowchart of a video search method according to an embodiment of the present application;

4 is another flowchart of a video search method according to an embodiment of the present application;

FIG. 5 is another flowchart of a video search method according to an embodiment of the present application;

FIG. 6 is another flowchart of a video search method according to an embodiment of the present application;

FIG. 7 is another flowchart of a video search method according to an embodiment of the present application;

FIG. 8 is another flowchart of a video search method according to an embodiment of the present application;

FIG. 9 is a schematic structural diagram of a cloud storage system according to an embodiment of the present application;

FIG. 10 is a schematic structural diagram of a video search apparatus according to an embodiment of the present application;

FIG. 11 is a schematic structural diagram of a computer device according to an embodiment of the present application.

Detailed ways

The technical solutions in the embodiments of the present application are clearly and completely described in the following with reference to the drawings in the embodiments of the present application. It is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present application without departing from the inventive scope are the scope of the present application.

The present application will be described in detail below through specific embodiments.

Please refer to FIG. 2, which shows a system architecture diagram of a method applying the embodiment of the present application. As shown in FIG. 2, the system to which the method of the embodiment of the present application is applied may include a cloud storage system 210 and a client 220. The cloud storage system 210 can include a management node 211 and a plurality of storage nodes 212. The storage node 212 stores a monitoring video. The user can search for the desired monitoring video in the storage node 212 through the client 220. Specifically, the client can send the video search request to the storage node 212 through the management node 211, and then perform video search in the storage node 212.

Specifically, please refer to FIG. 3, which illustrates a video search method process according to an embodiment of the present application, and the method may include the following steps:

S301. The management node receives a video search request that is sent by the client and includes feature information of the to-be-searched surveillance video, where the feature information includes at least one of the following: a reference picture, a spatial range, a time range, and attribute information.

In the embodiment of the present application, after the cloud storage system stores the monitoring video, the user may search for the monitoring video of interest in the cloud storage system, such as including all monitoring videos of a certain person, or include a certain vehicle in a certain period of time. All monitoring videos and more. Specifically, the user may generate a video search request including feature information of the to-be-searched video on the client, and send the video search request to the management node of the cloud storage system, and the management node may perform the search request according to the video search node. A monitored video that satisfies the condition is searched for in the stored monitoring video. The foregoing feature information may include at least one of the following: a reference picture, a spatial range, a time range, and attribute information.

For example, when searching for a surveillance video containing a certain person, the user can upload the photo of the person on the client as a reference picture, so that the cloud storage system can be searched for the surveillance video containing the person according to the photo; or When you want to search for a surveillance video of a person for a certain period of time, and know the person's dressing situation during that period of time, such as red blouse black pants, the user can enter the person's photo and dressing situation, dressing as Attribute information, so that the cloud storage system can be searched for the monitoring video that satisfies the condition according to the photo and attribute information.

S302. The management node determines a target storage node that stores the to-be-searched surveillance video, and sends the video search request to each of the target storage nodes.

Typically, a cloud storage system can include multiple storage nodes. Moreover, the monitoring video stored in the different storage nodes is different, and the management node stores the index information of the monitoring video stored by each storage node, where the index information includes at least the rough location information corresponding to the monitoring video stored in each storage node. Such as the province, city, region information.

In the embodiment of the present application, after receiving the video search request, the management node may first determine the target storage node that stores the monitoring video to be searched, and then send the video search request to each target storage node to perform in each target storage node. Monitor video search.

For example, when the video search request does not include the description information such as the spatial range, the time range, and the attribute information, and only includes the reference picture, the management node may determine all the storage nodes in the cloud storage system as the target storage node; when the video search The request includes related information included in the index information stored by the management node, such as a space range, etc., the management node may according to the spatial range included in the video search request, and the index information of the monitoring video stored in each locally stored storage node. Determine the target storage node where the monitored video to be searched is stored.

S303, each of the target storage nodes receives the video search request, and the matching degree between the search and the feature information is greater than the first according to the locally stored monitoring video, the stored monitoring video index information, and the feature information. The target of the preset threshold monitors the video index information, and sends the target monitoring video index information to the management node.

After receiving the video search request, each target storage node may search for the matching degree with the feature information according to the locally stored monitoring video, the stored monitoring video index information, and the feature information included in the video search request. The target of the preset threshold monitors the video index information, and sends the target monitoring video index information to the management node.

For example, any target storage node may sequentially read the monitoring video corresponding to each index information in all index information stored locally, and extract video feature information of the monitoring video. After that, the matching degree between the video feature information of the monitoring video and the feature information included in the video search request may be calculated. When the matching degree is greater than a preset threshold, such as 80%, 85%, or 90%, the index information is determined as a target. Monitor video index information.

In the embodiment of the present application, when the management node receives the video search request, it may first determine the target storage node storing the video to be searched, thereby transmitting the video search request to each target storage node, and each target storage node receives the video search request. After that, in the monitoring video stored by itself, the target monitoring video index information matching the video search request may be searched, and the searched target monitoring video index information is sent to the management node, so that the management node receives the The video search request matches all the target monitoring video index information, and realizes the search of the monitoring video in the cloud storage system. Moreover, each storage node separately searches for the target monitoring video index information stored by itself and matches the video search request, thereby ensuring the efficiency of monitoring the video search.

As an implementation manner of the embodiment of the present application, in order to improve the video search efficiency, when the feature information in the video search request received by the target storage node includes: a reference picture, a spatial range, a time range, and attribute information, each target storage The process of monitoring video search by the node, as shown in FIG. 4, may include the following steps:

S401. The target storage node performs feature extraction on the reference picture to obtain picture feature information corresponding to the reference picture.

In the embodiment of the present application, when the feature information includes a reference picture, a spatial range, a time range, and attribute information, any target storage node may perform a monitoring video search quickly and accurately according to the information.

Specifically, the target storage node may first perform feature extraction on the reference picture to obtain picture feature information corresponding to the reference picture. For example, the target storage node may perform feature extraction on the reference picture by using any feature extraction algorithm, which is not limited in this embodiment of the present application.

S402. The target storage node searches for the candidate monitoring video index information that matches the space range and the time range in the locally stored monitoring video index information.

In the embodiment of the present application, in order to improve the monitoring video search efficiency, the target storage node may not search in all the monitoring videos stored by itself, and may store the monitored video locally according to the spatial range and time range included in the feature information. The index information is used to find candidate monitoring video index information that matches the spatial extent and time range.

For example, the monitoring video index information stored in any target storage node can be as shown in Table 1:

Table 1

Surveillance video	Time information	location information
1	2017.3.2.1:00:00-2017.3.2.1:10:00	Street a
2	2017.3.2.1: 10:00-2017.3.2.1: 20:00	Street a
3	2017.3.2.1: 20:00-2017.3.2.1: 30:00	Street a

The above-mentioned monitoring video index information is merely illustrative of the content of the monitoring video index information stored in the storage node, and does not constitute a limitation on the monitoring video index information.

When the spatial range included in the feature information is street a, and the time range is 2017.3.2.1:01:00-2017.3.2.1:18:00, the storage node may find and check in the stored monitoring video index information. The candidate monitoring video index information matching the spatial range and the time range is the index information corresponding to the monitoring videos 1 and 2.

S403, the target storage node sequentially reads the monitoring video corresponding to each candidate monitoring video index information in the candidate monitoring video index information, and extracts video feature information of the monitoring video by using a pre-configured feature algorithm, according to the monitoring video. The video feature information, and the image feature information and the attribute information, calculate a matching degree between the candidate monitoring video index information and the feature information, and determine whether the matching degree is greater than a first preset threshold, and if so, The candidate monitoring video index information is used as the target monitoring video index information.

After determining the candidate monitoring video index information, the target storage node may further determine target monitoring video index information that satisfies the condition in the candidate monitoring video index information.

Specifically, the target storage node may sequentially read the monitoring video corresponding to each candidate monitoring video index information in the candidate monitoring video index information. For example, the target storage node may read the corresponding monitoring video from the corresponding location according to the storage location included in each candidate monitoring video index information. After the monitoring video is read, the storage node may further extract video feature information of the monitoring video.

In the embodiment of the present application, the monitoring video of the same application scenario is usually stored in each storage node. For example, one storage node stores road monitoring video, and another storage node stores cell monitoring video. It can be understood that the video features that are usually focused on for different scenes of surveillance video are also different. For example, for road surveillance video, it is usually concerned with vehicle characteristics; for community surveillance video, it is usually concerned with personnel characteristics.

Therefore, in the embodiment of the present application, a corresponding feature algorithm may be configured in each storage node in advance according to an application scenario of the monitoring video stored therein. For example, for a storage node storing a road monitoring video, a feature algorithm for performing feature extraction for the vehicle may be configured therein; and for a storage node storing the cell monitoring video, a feature algorithm for performing feature extraction for the person may be configured therein. When feature extraction is performed on the surveillance video, the storage node may perform feature extraction on the surveillance video according to a pre-configured feature algorithm, thereby obtaining video feature information of the surveillance video.

After obtaining the video feature information corresponding to any candidate monitoring video index information, the storage node may calculate the matching degree between the candidate monitoring video index information and the feature information according to the video feature information of the monitoring video, the image feature information and the attribute information. . For example, the storage node may calculate the matching degree between the candidate monitoring video index information and the feature information according to any matching degree calculation method, which is not limited in this embodiment of the present application.

After the matching of the candidate monitoring video index information and the feature information is calculated, the storage node may further determine whether the matching degree is greater than a first preset threshold, and if yes, use the candidate monitoring video index information as a target monitoring video index. information.

For example, when the target storage node determines that the index information corresponding to the monitoring video 1 and the monitoring video 2 is the candidate monitoring video index information, it may first determine whether the index information of the monitoring video 1 is the target monitoring video index information, and then determine the monitoring video 2 Whether the index information is the target monitoring video index information. Of course, it is also possible to determine whether the index information of the monitoring video 2 is the target monitoring video index information, and then determine whether the index information of the monitoring video 1 is the target monitoring video index information, which is not limited in this embodiment of the present application.

In this embodiment, the target storage node may first find the candidate monitoring video index information that satisfies the spatial range and the time range in the stored monitoring video index information, and then determine the target monitoring video index information, thereby improving the monitoring video search. effectiveness.

As an implementation manner of the embodiment of the present application, when the target storage node determines the target monitoring video index information in each candidate monitoring video index information, in order to improve accuracy, one or more candidate monitoring video index information is prevented from being missed, as shown in the figure. As shown in Figure 5, the process can include the following steps:

S501. The target storage node constructs an initial index information set including each candidate monitoring video index information, and constructs an empty result index information set.

In this embodiment, in order to improve the accuracy of the monitoring video search, after obtaining the candidate monitoring video index information, the target storage node may construct an initial index information set including each candidate monitoring video index information, and construct an empty result index information set. .

For example, when the target storage node determines that the candidate monitoring video index information is the index information corresponding to the monitoring video 1 and the monitoring video 2, it can construct the initial index information set as shown in Table 2:

Table 2

Surveillance video	Time information	location information
1	2017.3.2.1:00:00-2017.3.2.1:10:00	Street a
2	2017.3.2.1: 10:00-2017.3.2.1: 20:00	Street a

S502: Determine whether the initial index information set is empty; if not, execute step S503, and if yes, execute step S504.

S503. The monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the initial index information set, and the video feature information of the monitoring video is extracted by using a pre-configured feature algorithm, and according to the video of the monitoring video. The feature information, and the picture feature information and the attribute information, calculate a matching degree between the candidate monitoring video index information and the feature information, and determine whether the matching degree is greater than a first preset threshold, and if so, The candidate monitor video index information is added to the result index information set, and the candidate monitor video index information is deleted from the initial index information set, and the process returns to step S502.

When the target storage node determines that the initial index information set is not empty, the target storage node may sequentially read the monitoring video corresponding to the first candidate monitoring video index information in the initial index information set, and the monitoring video is Feature extraction is performed, and then the matching degree between the monitoring video index information and the feature information is determined according to the video feature information of the monitoring video, and the image feature information and the attribute information. The process is basically the same as the process of determining the matching degree of any candidate monitoring video index information and feature information in the embodiment shown in FIG. 4, and details are not described herein.

When the target storage node determines that the matching degree between the candidate monitoring video and the feature information is greater than the first preset threshold, the target monitoring video index information may be added to the result index information set, and the candidate monitoring video index information is initialized from the initial The index information set is deleted, and it is returned to determine whether the initial index information set is empty.

For example, when the initial index information set is as shown in Table 2, the target storage node may first determine, for the monitoring video 1, whether the matching degree of the index information of the monitoring video 1 and the feature information is greater than a first preset threshold, and if so, monitor The index information of the video 1 is added to the result index information set; if not, the index information of the monitoring video 1 is deleted from the initial index information set, and the judgment of the monitoring video 2 is continued to determine whether the index information of the monitoring video 2 is accessed. In the information collection.

S504. Use index information included in the result index information set as the target monitoring video index information.

When the initial index information set is empty, it indicates that all the candidate monitoring video index information has been determined. At this time, the index information in the result index information set is that the matching degree with the feature information is greater than the first preset threshold. Index information. In this case, the target storage node may directly use the index information included in the result index information set as the target monitoring video index information.

In this embodiment, the target monitoring video index information can be determined by constructing the initial index information set and the result index information set, thereby ensuring that each candidate monitoring video index information is judged and improving the accuracy of the monitoring video search.

Understandably, in some cases, different types of video searches may be performed for the same application scenario. For example, for community surveillance video, in addition to personnel search, sometimes vehicle search may be required.

Therefore, as an implementation manner of the embodiment of the present application, in order to improve the applicability and accuracy of the surveillance video search, when configuring the feature algorithm in each storage node, multiple feature algorithms may be configured in each storage node. Also, each feature algorithm can have its priority. For example, for a storage node storing a cell monitoring video, two feature algorithms may be configured therein, namely, a feature algorithm for feature extraction for a vehicle and a feature algorithm for feature extraction for a person, and a feature algorithm for feature extraction for a person The priority is higher than the priority of the feature algorithm for feature extraction of the vehicle.

Correspondingly, when the target storage node determines the target monitoring video index information in each candidate monitoring video index information, as shown in FIG. 6, the process may include the following steps:

S601. The target storage node constructs an initial index information set including each candidate monitoring video index information, and constructs an empty result index information set.

S602. Determine whether the initial index information set is empty. If not, execute step S603, and if yes, perform step S604.

S603. The monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the initial index information set, and the video feature information of the monitoring video is extracted by using a feature algorithm with the highest priority, and according to the monitoring video. The video feature information, and the image feature information and the attribute information, calculate a matching degree between the candidate monitoring video index information and the feature information, and determine whether the matching degree is greater than a first preset threshold, and if yes, The candidate monitoring video index information is added to the result index information set, and the candidate monitoring video index information is deleted from the initial index information set, and the process returns to step S602.

The above steps S601-S603 are basically the same as the steps S501-S503 in the embodiment shown in FIG. 5, and details are not described herein. The difference is that in step S503, when a plurality of feature algorithms are pre-configured in the target storage node, when the target storage node extracts video feature information of the surveillance video, the feature algorithm with the highest priority is used.

S604. Determine whether the result index information set is empty. If yes, go to step S605, if no, go to step S608.

In the embodiment of the present application, after obtaining the result index information set, the target storage node may determine whether the result index information set is empty.

When the target storage node determines that the result index information set is empty, it may be caused by a mistake in the feature algorithm used. In this case, the target storage node may perform feature extraction on the monitoring video by using other feature algorithms, and re-determine the target monitoring video feature information in the candidate monitoring video index information according to the extracted video feature information.

S605. The target storage node again constructs an initial index information set including each candidate monitoring video index information.

S606. Determine whether the newly constructed initial index information set is empty. If not, execute step S607, and if yes, return to step S604.

S607. The monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the newly constructed initial index information set, and the video feature information of the monitoring video is extracted by using a next priority feature algorithm, and according to The video feature information of the monitoring video, and the image feature information and the attribute information, calculating a matching degree between the candidate monitoring video index information and the feature information, and determining whether the matching degree is greater than a first preset threshold. If yes, the candidate monitoring video index information is added to the result index information set, and the candidate monitoring video index information is deleted from the newly constructed initial index information set, and the process returns to step S606.

The above steps S605-S607 are substantially the same as the above steps S601-S603, and are not described herein. The difference is that in step S607, when the target storage node extracts the video feature information of the surveillance video, the feature algorithm of the next priority is used.

That is, the target storage node performs feature extraction on the monitoring video by using the feature algorithm with the highest priority in step S603, and performs the matching degree calculation, and when the obtained result index information set is empty, the priority is only used in step S607. The feature algorithm next to the highest priority again performs feature extraction on the surveillance video and performs matching calculation.

After feature extraction is performed on the monitoring video corresponding to the candidate monitoring video index information by using the feature algorithm with the priority next to the highest priority, and the matching degree is calculated, the result index information set may be judged again to be empty, and if so, it is used again. The feature algorithm of the third priority performs feature extraction on the monitoring video corresponding to all candidate monitoring video index information and performs matching degree calculation, and loops the above process until the result index information set is not empty, or all feature algorithm pairs are used. Monitor the video corresponding to the video index information for feature extraction and perform matching calculation.

S608. Use index information included in the result index information set as the target monitoring video index information.

When the storage node determines that the result index information set is not empty, it may use the index information included in the result index information set as the target monitoring video index information.

In this embodiment, the feature video can be extracted by using multiple feature algorithms, thereby improving the comprehensiveness and accuracy of the surveillance video search.

As an implementation manner of the embodiment of the present application, after receiving the target monitoring video index information sent by each target storage node, the management node may send the target monitoring video index information to the client. Moreover, the management node may further adjust the first preset threshold in each target storage node according to the feedback information of the client.

Specifically, after receiving the target monitoring video index information sent by the management node, the client may obtain corresponding monitoring video from each target storage node according to the target monitoring video index information, and display, so that the user can view the displayed on the client. Monitor the video. For example, the client may obtain and display each monitoring video from the corresponding location of the corresponding storage node according to the storage node identifier and storage location of each monitoring video included in the monitoring video index information.

After the user views the displayed surveillance video on the client, it can also determine whether the surveillance video is the surveillance video that it wants to search for and input confirmation information. For example, it can input the accuracy of the surveillance video, that is, it can confirm the proportion of the surveillance video that meets the requirements of the monitored video in the searched video. After receiving the confirmation information input by the user, the client may send the confirmation information to the management node.

After receiving the confirmation information, the management node may adjust the first preset threshold in each target storage node according to the confirmation information. For example, when the accuracy rate included in the confirmation information is greater than the second preset threshold, it indicates that the current first preset threshold is relatively large; in this case, the management node may decrease the first preset threshold to search for more. Monitoring video. When the accuracy rate included in the confirmation information is less than or equal to the second threshold, it indicates that the current first preset threshold is relatively small, and the monitoring video search accuracy is low; in this case, the management node may increase the first preset threshold. To improve the accuracy of surveillance video search.

In this embodiment, the management node may adjust the first preset threshold of the target storage node according to the feedback information of the client, so that when the monitoring video search is performed again, more monitoring videos satisfying the requirements may be searched, or the monitoring may be improved. The accuracy of video search.

As an implementation manner of the embodiment of the present application, after the management node determines the target storage node, in order to accurately send the video search request to each target storage node accurately, as shown in FIG. 7, the process may include the following steps:

S701. The management node constructs a target storage node list that includes each target storage node identifier.

In the embodiment of the present application, after determining the target storage node, the management node may construct a target storage node list including the identifiers of the target storage nodes. For example, the target storage node list constructed by the management node may include a target storage node 1, a target storage node 2, and a target storage node 3.

S702. Determine whether the target storage node list is empty. If not, execute step S703, and if yes, end.

S703. Send the video search request to a target storage node in the target storage node list, and determine whether the transmission is successful. If yes, go to step S704, and if no, go back to step S702.

When the management node determines that the target storage node list is not empty, it may send a video search request to a target storage node in the target storage node list. After sending the video search request to a target storage node, the management node can also determine if the transmission was successful.

For example, the management node may send the video search request to the top target storage node according to the order of the target storage nodes in the target storage node list. When the storage node 1, the storage node 2, and the storage node 3 are sequentially included in the target storage node list, the management node may first send a video search request to the target storage node 1.

S704. The target storage node identifier is removed from the target storage node list, and the process returns to step S702.

After the management node sends the video search request to a target storage node and confirms that the transmission is successful, the management node may remove the target storage node identifier from the target storage node list, and return to determine whether the target storage node list is empty, to ensure Send video search requests to all target storage nodes.

In this embodiment, the management node may send the video search request to the target storage node by constructing the target storage node list, so that the video search request can be successfully sent to each target storage node.

As an implementation manner of the embodiment of the present application, before performing the monitoring video search, the client may first construct a video search request. Specifically, when performing a surveillance video search, the user may upload a reference picture on the client. For example, it can upload a picture containing a person or a car as a reference picture.

When the user wants to search for a surveillance video of a person for a period of time, and probably knows the location of the person's space during that time, and the person's dressing situation, the user can enter this information into the client. For example, the user can input all the information he knows in the input interface provided by the client to improve the accuracy of the video search.

After receiving the reference picture, spatial range, time range and attribute information input by the user, the client may integrate the reference picture, the spatial range, the time range, and the attribute information according to a predetermined format standard to generate a video search request, and The video search request is sent to the management node to cause the management node to perform a surveillance video search based on the video search request.

In this embodiment, the client can generate a video search request that meets the requirements, thereby ensuring that the cloud storage system can successfully perform the surveillance video search.

Correspondingly, the embodiment of the present application further provides a video search method, which is applied to a target storage node in a cloud storage system, where the cloud storage system further includes a management node. As shown in FIG. 8, the method includes the following steps:

S801. Receive a video search request that is sent by the management node and that includes the feature information of the to-be-searched video, where the video search request is that the management node determines that the target storage node stores the to-be-searched video. The feature information sent by the target storage node includes at least one of the following: a reference picture, a spatial range, a time range, and attribute information.

In the embodiment of the present application, after the cloud storage system stores the monitoring video, the user may search for the monitoring video of interest in the cloud storage system, such as including all monitoring videos of a certain person, or include a certain vehicle in a certain period of time. All monitoring videos and more. Specifically, the user may generate a video search request including feature information of the to-be-searched video on the client, and send the video search request to the management node of the cloud storage system, and the management node may perform the search request according to the video search node. A monitored video that satisfies the condition is searched for in the stored monitoring video. The foregoing feature information may include at least one of the following: a reference picture, a spatial range, a time range, and attribute information.

After receiving the video search request, the management node may first determine the target storage node storing the monitoring video to be searched, and then send the video search request to each target storage node to perform monitoring video search in each target storage node.

For example, when the video search request does not include the description information such as the spatial range, the time range, and the attribute information, and only includes the reference picture, the management node may determine all the storage nodes in the cloud storage system as the target storage node; when the video search The request includes related information included in the index information stored by the management node, such as a space range, etc., the management node may according to the spatial range included in the video search request, and the index information of the monitoring video stored in each locally stored storage node. Determine the target storage node where the monitored video to be searched is stored.

S802: Search, according to the locally stored monitoring video, the stored monitoring video index information, and the feature information, a target monitoring video index information that is greater than a first preset threshold between the feature information and the target information. The target monitoring video index information is sent to the management node.

After receiving the video search request, the target storage node may search for the matching degree with the feature information according to the locally stored monitoring video, the stored monitoring video index information, and the feature information included in the video search request. The target monitoring video index information is set, and the target monitoring video index information is sent to the management node.

For example, the target storage node may sequentially read the monitoring video corresponding to each index information in all the index information stored locally, and extract the video feature information of the monitoring video. After that, the matching degree between the video feature information of the monitoring video and the feature information included in the video search request may be calculated. When the matching degree is greater than a preset threshold, such as 80%, 85%, or 90%, the index information is determined as a target. Monitor video index information.

In the embodiment of the present application, when the management node receives the video search request, it may first determine the target storage node storing the video to be searched, thereby transmitting the video search request to each target storage node, and each target storage node receives the video search request. After that, in the monitoring video stored by itself, the target monitoring video index information matching the video search request may be searched, and the searched target monitoring video index information is sent to the management node, so that the management node receives the The video search request matches all the target monitoring video index information, and realizes the search of the monitoring video in the cloud storage system. Moreover, each storage node separately searches for the target monitoring video index information stored by itself and matches the video search request, thereby ensuring the efficiency of monitoring the video search.

As an implementation manner of the embodiment of the present application, in order to improve the video search efficiency, when the feature information in the video search request received by the target storage node includes: a reference picture, a spatial range, a time range, and attribute information, each target storage The process by which a node performs a surveillance video search may include:

Performing feature extraction on the reference picture to obtain picture feature information corresponding to the reference picture;

Finding candidate monitoring video index information that matches the spatial range and time range in the locally stored monitoring video index information;

In the candidate monitoring video index information, the monitoring video corresponding to each candidate monitoring video index information is sequentially read, and the video feature information of the monitoring video is extracted by using a pre-configured feature algorithm, according to the video feature information of the monitoring video, and Determining, by the picture feature information and the attribute information, a matching degree between the candidate monitoring video index information and the feature information, and determining whether the matching degree is greater than a first preset threshold, and if yes, selecting the candidate monitoring video The index information is used as the target monitoring video index information.

In this embodiment, the target storage node may first find the candidate monitoring video index information that satisfies the spatial range and the time range in the stored monitoring video index information, and then determine the target monitoring video index information, thereby improving the monitoring video search. effectiveness.

As an implementation manner of the embodiment of the present application, when the target storage node determines the target monitoring video index information in each candidate monitoring video index information, in order to improve accuracy and prevent one or more candidate monitoring video index information from being missed, the process The steps can be included:

Constructing an initial index information set including each candidate monitoring video index information, and constructing an empty result index information set;

Determining whether the initial index information set is empty;

If not, the monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the initial index information set, and the video feature information of the monitoring video is extracted by using a pre-configured feature algorithm, and according to the monitoring video The video feature information, and the image feature information and the attribute information, calculate a matching degree between the candidate monitoring video index information and the feature information, and determine whether the matching degree is greater than a first preset threshold, and if yes, Adding the candidate monitoring video index information to the result index information set, deleting the candidate monitoring video index information from the initial index information set, and returning to perform the step of determining whether the initial index information set is empty;

If yes, the index information included in the result index information set is used as the target monitoring video index information.

In this embodiment, the target monitoring video index information can be determined by constructing the initial index information set and the result index information set, thereby ensuring that each candidate monitoring video index information is judged and improving the accuracy of the monitoring video search.

Understandably, in some cases, different types of video searches may be performed for the same application scenario. For example, for community surveillance video, in addition to conducting personnel searches, vehicle searches may sometimes be required.

Therefore, as an implementation manner of the embodiment of the present application, in order to improve the applicability and accuracy of the surveillance video search, when configuring the feature algorithm in each storage node, multiple feature algorithms may be configured in each storage node. Also, each feature algorithm can have its priority. For example, for a storage node storing a cell monitoring video, two feature algorithms may be configured therein, namely, a feature algorithm for feature extraction for a vehicle and a feature algorithm for feature extraction for a person, and a feature algorithm for feature extraction for a person The priority is higher than the priority of the feature algorithm for feature extraction of the vehicle.

Correspondingly, when the target storage node determines the target monitoring video index information in each candidate monitoring video index information, the process may include the following steps:

Extracting video feature information of the surveillance video by using a feature algorithm with the highest priority;

The step of using the index information included in the result index information set as the target monitoring video index information includes:

Determining whether the result index information set is empty; when the result index information set is empty, the target storage node again constructs an initial index information set including each candidate monitoring video index information, and determines the newly constructed initial index information set. Whether it is empty;

When the newly constructed initial index information set is not empty, the monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the newly constructed initial index information set, and the next priority feature algorithm is adopted. Extracting video feature information of the monitoring video, and calculating a matching degree between the candidate monitoring video index information and the feature information according to the video feature information of the monitoring video, the image feature information, and the attribute information, and determining If the matching degree is greater than the first preset threshold, if yes, the candidate monitoring video index information is added to the result index information set, and the candidate monitoring video index information is deleted from the newly constructed initial index information set. And returning to the step of performing the determining whether the newly constructed initial index information set is empty;

When the newly constructed initial index information set is empty, returning to perform the step of determining whether the result index information set is empty;

When the result index information set is not empty, the index information included in the result index information set is used as the target monitoring video index information.

In this embodiment, the feature video can be extracted by using multiple feature algorithms, thereby improving the comprehensiveness and accuracy of the surveillance video search.

Correspondingly, the embodiment of the present application further provides a cloud storage system. As shown in FIG. 9, the cloud storage system includes: a management node 910 and a plurality of storage nodes 920.

The management node 910 is configured to receive a video search request that is sent by the client and includes feature information of the to-be-searched surveillance video, where the feature information includes at least one of the following: a reference picture, a spatial range, a time range, and attribute information. Determining a target storage node 920 storing the to-be-searched video, and transmitting the video search request to each of the target storage nodes 920;

Each of the target storage nodes 920 is configured to receive the video search request, and according to the locally stored monitoring video, the stored monitoring video index information, and the feature information, the matching degree between the search and the feature information is greater than The target of the first preset threshold monitors video index information, and sends the target monitoring video index information to the management node 910.

In the embodiment of the present application, when the management node receives the video search request, it may first determine the target storage node storing the video to be searched, thereby transmitting the video search request to each target storage node, and each target storage node receives the video search request. After that, in the monitoring video stored by itself, the target monitoring video index information matching the video search request may be searched, and the searched target monitoring video index information is sent to the management node, so that the management node receives the The video search request matches all the target monitoring video index information, and realizes the search of the monitoring video in the cloud storage system. Moreover, each storage node separately searches for the target monitoring video index information stored by itself and matches the video search request, thereby ensuring the efficiency of monitoring the video search.

As an implementation manner of the embodiment of the present application, when the feature information includes: a reference picture, a spatial range, a time range, and attribute information,

Any target storage node 920 is specifically configured to perform feature extraction on the reference picture to obtain picture feature information corresponding to the reference picture; and search for the spatial range and time range in the locally stored monitoring video index information. The candidate monitoring video index information; in the candidate monitoring video index information, the monitoring video corresponding to each candidate monitoring video index information is sequentially read, and the video feature information of the monitoring video is extracted by using a pre-configured feature algorithm, according to the monitoring video The video feature information, and the image feature information and the attribute information, calculate a matching degree between the candidate monitoring video index information and the feature information, and determine whether the matching degree is greater than a first preset threshold, and if so, The candidate monitoring video index information is used as the target monitoring video index information.

As an implementation manner of the embodiment of the present application, any target storage node 920 is specifically configured to construct an initial index information set including each candidate monitoring video index information, and construct an empty result index information set; and determine the initial index information. Whether the set is empty; if not, the monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the initial index information set, and the video feature information of the monitoring video is extracted by using a pre-configured feature algorithm, and Calculating, according to the video feature information of the monitoring video, the image feature information and the attribute information, a matching degree between the candidate monitoring video index information and the feature information, and determining whether the matching degree is greater than a first preset threshold If yes, add the candidate monitoring video index information to the result index information set, delete the candidate monitoring video index information from the initial index information set, and return to perform the determining whether the initial index information set is Empty step; if yes, the result index information The index information included in the set is used as the target monitoring video index information.

As an implementation manner of the embodiment of the present application, multiple feature algorithms are pre-configured in each storage node, and each feature algorithm has its corresponding priority.

Any target storage node 920 is specifically configured to extract video feature information of the monitoring video by using a feature algorithm with the highest priority; and when the index information included in the result index information set is used as the target monitoring video index information, Whether the result index information set is empty; when the result index information set is empty, the target storage node again constructs an initial index information set including each candidate monitoring video index information, and determines whether the newly constructed initial index information set is Is empty;

When the newly constructed initial index information set is not empty, the monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the newly constructed initial index information set, and the next priority feature algorithm is adopted. Extracting video feature information of the monitoring video, and calculating a matching degree between the candidate monitoring video index information and the feature information according to the video feature information of the monitoring video, the image feature information, and the attribute information, and determining If the matching degree is greater than the first preset threshold, if yes, the candidate monitoring video index information is added to the result index information set, and the candidate monitoring video index information is deleted from the newly constructed initial index information set. And returning to the step of performing the determining whether the newly constructed initial index information set is empty;

When the newly constructed initial index information set is empty, returning to perform the step of determining whether the result index information set is empty;

When the result index information set is not empty, the index information included in the result index information set is used as the target monitoring video index information.

As an implementation manner of the embodiment of the present application, the management node 910 is further configured to receive target monitoring video index information sent by each target storage node, and send the target monitoring video index information to the client.

As an implementation manner of the embodiment of the present application, the management node 910 is further configured to receive the acknowledgement information sent by the client, and adjust the first preset threshold in each target storage node according to the acknowledgement information.

As an implementation manner of the embodiment of the present application, the management node 910 is specifically configured to: when the accuracy rate included in the confirmation information is greater than a second preset threshold, reduce the first preset threshold; When the accuracy rate included in the confirmation information is less than or equal to the second threshold, the first preset threshold is increased.

As an implementation manner of the embodiment of the present application, the management node 910 is specifically configured to determine, according to the spatial range, and the index information of the monitoring video stored in each locally stored storage node, that the to-be-searched monitoring is stored. The target storage node for the video.

As an implementation manner of the embodiment of the present application, the management node 910 is specifically configured to construct a target storage node list that includes each target storage node identifier; determine whether the target storage node list is empty; and when the target storage node is When the list is not empty, sending the video search request to a target storage node in the target storage node list, and determining whether the sending is successful; when the sending is successful, identifying the target storage node identifier from the target storage node list And removing, performing the step of determining whether the target storage node list is empty; when the sending fails, returning to performing the step of determining whether the storage node list is empty.

Correspondingly, the embodiment of the present application further provides a video search device, which is applied to a target storage node in a cloud storage system, where the cloud storage system further includes a management node. As shown in FIG. 10, the device includes:

The receiving module 1010 is configured to receive, by the management node, a video search request that includes feature information of a to-be-searched surveillance video, where the video search request is that the management node determines that the target storage node stores a to-be-searched surveillance video. Sending to the target storage node, the feature information includes at least one of the following: a reference picture, a spatial range, a time range, and attribute information;

The search module 1020 is configured to search, according to the locally stored monitoring video, the stored monitoring video index information, and the feature information, the target monitoring video index information that is matched with the feature information and is greater than the first preset threshold. And sending the target monitoring video index information to the management node.

In the embodiment of the present application, when the management node receives the video search request, it may first determine the target storage node storing the video to be searched, thereby transmitting the video search request to each target storage node, and each target storage node receives the video search request. After that, in the monitoring video stored by itself, the target monitoring video index information matching the video search request may be searched, and the searched target monitoring video index information is sent to the management node, so that the management node receives the The video search request matches all the target monitoring video index information, and realizes the search of the monitoring video in the cloud storage system. Moreover, each storage node separately searches for the target monitoring video index information stored by itself and matches the video search request, thereby ensuring the efficiency of monitoring the video search.

As an implementation manner of the embodiment of the present application, when the feature information includes: a reference picture, a spatial range, a time range, and attribute information, the searching module 1020 includes:

a feature extraction sub-module (not shown) for performing feature extraction on the reference picture to obtain picture feature information corresponding to the reference picture;

a sub-module (not shown) for searching for candidate monitoring video index information matching the spatial range and time range in the locally stored monitoring video index information;

The search sub-module (not shown) is configured to sequentially read the monitoring video corresponding to each candidate monitoring video index information in the candidate monitoring video index information, and extract the video of the monitoring video by using a pre-configured feature algorithm. Feature information, according to the video feature information of the monitoring video, and the image feature information and the attribute information, calculating a matching degree between the candidate monitoring video index information and the feature information, and determining whether the matching degree is greater than the first The preset threshold, if yes, the candidate monitoring video index information is used as the target monitoring video index information.

As an implementation manner of the embodiment of the present application, the search submodule includes:

Constructing a subunit (not shown) for constructing an initial index information set including each candidate monitoring video index information, and constructing an empty result index information set;

a determining subunit (not shown) for determining whether the initial index information set is empty;

a feature extraction sub-unit (not shown), configured to sequentially read, in the initial index information set, a monitoring corresponding to a piece of candidate monitoring video index information located in the front, when the determining sub-unit determines that the result is negative Video, extracting video feature information of the surveillance video by using a pre-configured feature algorithm;

Searching a subunit (not shown) for calculating a matching degree between the candidate monitoring video index information and the feature information according to the video feature information of the monitoring video, and the image feature information and the attribute information And determining whether the matching degree is greater than the first preset threshold, and if yes, adding the candidate monitoring video index information to the result index information set, and deleting the candidate monitoring video index information from the initial index information set, And triggering the determining subunit;

Determining a subunit (not shown) for using the index information included in the result index information set as the target monitoring video index information when the judgment subunit determines that the result is YES.

As an implementation manner of the embodiment of the present application, the feature extraction sub-unit is specifically configured to extract video feature information of the surveillance video by using a feature algorithm with the highest priority;

The determining subunit is specifically configured to:

Determining whether the result index information set is empty; when the result index information set is empty, the target storage node again constructs an initial index information set including each candidate monitoring video index information, and determines the newly constructed initial index information set. Whether it is empty;

When the newly constructed initial index information set is not empty, the monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the newly constructed initial index information set, and the next priority feature algorithm is adopted. Extracting video feature information of the monitoring video, and calculating a matching degree between the candidate monitoring video index information and the feature information according to the video feature information of the monitoring video, the image feature information, and the attribute information, and determining If the matching degree is greater than the first preset threshold, if yes, the candidate monitoring video index information is added to the result index information set, and the candidate monitoring video index information is deleted from the newly constructed initial index information set. And returning to the step of performing the determining whether the newly constructed initial index information set is empty;

When the newly constructed initial index information set is empty, returning to perform the step of determining whether the result index information set is empty;

When the result index information set is not empty, the index information included in the result index information set is used as the target monitoring video index information.

Correspondingly, the embodiment of the present application further provides a computer device, as shown in FIG. 11, the computer device includes: a processor 1110, a communication interface 1120, a memory 1130, and a communication bus 1140, wherein the processor 1110, the The communication interface 1120 and the memory 1130 complete communication with each other through the communication bus 1140;

The memory 1130 is configured to store a computer program;

The processor 1110 is configured to implement a video search method applied to a target storage node in a cloud storage system according to the foregoing embodiment when the program stored in the memory 1130 is executed.

In the embodiment of the present application, when the management node receives the video search request, it may first determine the target storage node storing the video to be searched, thereby transmitting the video search request to each target storage node, and each target storage node receives the video search request. After that, in the monitoring video stored by itself, the target monitoring video index information matching the video search request may be searched, and the searched target monitoring video index information is sent to the management node, so that the management node receives the The video search request matches all the target monitoring video index information, and realizes the search of the monitoring video in the cloud storage system. Moreover, each storage node separately searches for the target monitoring video index information stored by itself and matches the video search request, thereby ensuring the efficiency of monitoring the video search.

The communication bus 1140 mentioned in the above computer device may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The communication bus 1140 can be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, only one line is shown in the figure, but it does not mean that there is only one bus or one type of bus.

Communication interface 1120 is used for communication between the aforementioned computer devices and other devices.

The memory 1130 may include a random access memory (RAM), and may also include a non-volatile memory, such as at least one disk storage. Optionally, the memory 1130 may also be at least one storage device located away from the foregoing processor.

The above-mentioned processor 1110 may be a general-purpose processor, including a central processing unit (CPU), a network processor (Ne twork processor, NP for short), and the like; or a digital signal processor (Digital Signal Processing, referred to as DSP), Application Specific Integrated Circuit (ASIC), Field-Programmable Gate Array (FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware component.

Correspondingly, the embodiment of the present application further provides a computer readable storage medium, where the computer readable storage medium stores a computer program, and when the computer program is executed by the processor, the application is implemented as described in the foregoing embodiment. A video search method for a target storage node in a cloud storage system.

In the embodiment of the present application, when the management node receives the video search request, it may first determine the target storage node storing the video to be searched, thereby transmitting the video search request to each target storage node, and each target storage node receives the video search request. After that, in the monitoring video stored by itself, the target monitoring video index information matching the video search request may be searched, and the searched target monitoring video index information is sent to the management node, so that the management node receives the The video search request matches all the target monitoring video index information, and realizes the search of the monitoring video in the cloud storage system. Moreover, each storage node separately searches for the target monitoring video index information stored by itself and matches the video search request, thereby ensuring the efficiency of monitoring the video search.

Correspondingly, the embodiment of the present application further provides an application program, where the application is used to execute a video search method applied to a target storage node in a cloud storage system according to an embodiment of the present application at runtime.

In the embodiment of the present application, when the management node receives the video search request, it may first determine the target storage node storing the video to be searched, thereby transmitting the video search request to each target storage node, and each target storage node receives the video search request. After that, in the monitoring video stored by itself, the target monitoring video index information matching the video search request may be searched, and the searched target monitoring video index information is sent to the management node, so that the management node receives the The video search request matches all the target monitoring video index information, and realizes the search of the monitoring video in the cloud storage system. Moreover, each storage node separately searches for the target monitoring video index information stored by itself and matches the video search request, thereby ensuring the efficiency of monitoring the video search.

For the device/cloud storage system/computer device/storage medium/application embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method embodiment.

It should be noted that, in this context, relational terms such as first and second are used merely to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply these entities or operations. There is any such actual relationship or order between them. Furthermore, the term "comprises" or "comprises" or "comprises" or any other variations thereof is intended to encompass a non-exclusive inclusion, such that a process, method, article, or device that comprises a plurality of elements includes not only those elements but also Other elements, or elements that are inherent to such a process, method, item, or device. An element that is defined by the phrase "comprising a ..." does not exclude the presence of additional equivalent elements in the process, method, item, or device that comprises the element.

The various embodiments in the present specification are described in a related manner, and the same or similar parts between the various embodiments may be referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method embodiment.

The above description is only the preferred embodiment of the present application, and is not intended to limit the scope of the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application are included in the scope of the present application.

Claims (29)

1. A video search method, which is applied to a cloud storage system, where the cloud storage system includes a management node and a plurality of storage nodes, and the method includes:

   The management node receives a video search request that is sent by the client and includes feature information of the to-be-searched surveillance video, where the feature information includes at least one of the following: a reference picture, a spatial range, a time range, and attribute information;

   The management node determines a target storage node that stores the to-be-searched surveillance video, and sends the video search request to each of the target storage nodes;

   Each of the target storage nodes receives the video search request, and according to the locally stored monitoring video, the stored monitoring video index information, and the feature information, the matching degree between the search and the feature information is greater than the first preset. The target of the threshold monitors the video index information, and sends the target monitoring video index information to the management node.
2. The method according to claim 1, wherein when the feature information comprises: a reference picture, a spatial range, a time range, and attribute information, each of the target storage nodes is based on a locally stored monitoring video, The step of storing the monitoring video index information and the feature information, and searching for the target monitoring video index information that is greater than the first preset threshold by the matching of the feature information includes:

   Performing feature extraction on the reference picture by any target storage node to obtain picture feature information corresponding to the reference picture;

   The target storage node searches for the candidate monitoring video index information that matches the spatial range and the time range in the locally stored monitoring video index information;

   The target storage node sequentially reads the monitoring video corresponding to each candidate monitoring video index information in the candidate monitoring video index information, and extracts video feature information of the monitoring video by using a pre-configured feature algorithm, according to the video of the monitoring video. The feature information, and the picture feature information and the attribute information, calculate a matching degree between the candidate monitoring video index information and the feature information, and determine whether the matching degree is greater than a first preset threshold, and if so, The candidate monitor video index information is used as the target monitoring video index information.
3. The method according to claim 2, wherein the target storage node sequentially reads the monitoring video corresponding to each candidate monitoring video index information in the candidate monitoring video index information, and adopts a pre-configured feature algorithm. Extracting video feature information of the monitoring video, calculating a matching degree between the candidate monitoring video index information and the feature information according to the video feature information of the monitoring video, and the image feature information and the attribute information, and determining the Whether the matching degree is greater than the first preset threshold, if yes, the step of using the candidate monitoring video index information as the target monitoring video index information includes:

   The target storage node constructs an initial index information set including each candidate monitoring video index information, and constructs an empty result index information set;

   Determining whether the initial index information set is empty;

   If not, the monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the initial index information set, and the video feature information of the monitoring video is extracted by using a pre-configured feature algorithm, and according to the monitoring video The video feature information, and the image feature information and the attribute information, calculate a matching degree between the candidate monitoring video index information and the feature information, and determine whether the matching degree is greater than a first preset threshold, and if yes, Adding the candidate monitoring video index information to the result index information set, deleting the candidate monitoring video index information from the initial index information set, and returning to perform the step of determining whether the initial index information set is empty;

   If yes, the index information included in the result index information set is used as the target monitoring video index information.
4. The method according to claim 3, wherein each of the storage nodes is pre-configured with a plurality of feature algorithms, and each feature algorithm has a corresponding priority, and the pre-configured feature algorithm is used to extract the video of the surveillance video. The steps of the feature information include:

   Extracting video feature information of the surveillance video by using a feature algorithm with the highest priority;

   The step of using the index information included in the result index information set as the target monitoring video index information includes:

   Determining whether the result index information set is empty; when the result index information set is empty, the target storage node again constructs an initial index information set including each candidate monitoring video index information, and determines the newly constructed initial index information set. Whether it is empty; when the result index information set is not empty, the index information included in the result index information set is used as the target monitoring video index information;

   When the newly constructed initial index information set is not empty, the monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the newly constructed initial index information set, and the next priority feature algorithm is adopted. Extracting video feature information of the monitoring video, and calculating a matching degree between the candidate monitoring video index information and the feature information according to the video feature information of the monitoring video, the image feature information, and the attribute information, and determining If the matching degree is greater than the first preset threshold, if yes, the candidate monitoring video index information is added to the result index information set, and the candidate monitoring video index information is deleted from the newly constructed initial index information set. And returning to the step of performing the determining whether the newly constructed initial index information set is empty;

   When the newly constructed initial index information set is empty, a step of performing the determination as to whether the result index information set is empty is returned.
5. The method according to claim 3 or 4, wherein after the sending the target monitoring video index information to the management node, the method further comprises:

   The management node receives target monitoring video index information sent by each target storage node, and sends the target monitoring video index information to the client.
6. The method according to claim 5, wherein after the sending the target monitoring video index information to the client, the method further comprises:

   The management node receives the confirmation information sent by the client, and adjusts a first preset threshold in each target storage node according to the confirmation information.
7. The method according to claim 6, wherein the step of adjusting the first preset threshold in each target storage node according to the confirmation information comprises:

   When the accuracy rate included in the confirmation information is greater than a second preset threshold, decreasing the first preset threshold;

   When the accuracy rate included in the confirmation information is less than or equal to the second threshold, the first preset threshold is increased.
8. The method according to any one of claims 2 to 5, wherein the determining, by the management node, the target storage node storing the monitoring video to be searched comprises:

   The management node determines, according to the spatial range and the index information of the monitoring video stored in each locally stored storage node, the target storage node that stores the to-be-searched surveillance video.
9. The method according to any one of claims 1-5, wherein the step of transmitting the video search request to each of the target storage nodes comprises:

   The management node constructs a list of target storage nodes including identifiers of the target storage nodes;

   Determining whether the target storage node list is empty; when the target storage node list is not empty, sending the video search request to a target storage node in the target storage node list, and determining whether the transmission is successful;

   When the sending is successful, removing the target storage node identifier from the target storage node list, and returning to perform the step of determining whether the target storage node list is empty;

   When the transmission fails, a step of performing the determination of whether the storage node list is empty is returned.
10. A video search method, which is applied to a target storage node in a cloud storage system, the cloud storage system further includes a management node, and the method includes:

    Receiving, by the management node, a video search request that includes feature information of a to-be-searched surveillance video, where the video search request is that the management node determines that the target storage node stores a to-be-searched surveillance video, and stores the target information The feature information sent by the node includes at least one of the following: a reference picture, a spatial range, a time range, and attribute information;

    Searching, according to the locally stored monitoring video, the stored monitoring video index information, and the feature information, target monitoring video index information that is greater than a first preset threshold between the feature information and the target The monitoring video index information is sent to the management node.
11. The method according to claim 10, wherein when the feature information comprises: a reference picture, a spatial range, a time range, and attribute information, the locally stored monitoring video and the stored monitoring video index information And the step of searching for the target monitoring video index information that the matching degree with the feature information is greater than the first preset threshold includes:

    Performing feature extraction on the reference picture to obtain picture feature information corresponding to the reference picture;

    Finding candidate monitoring video index information that matches the spatial range and time range in the locally stored monitoring video index information;

    In the candidate monitoring video index information, the monitoring video corresponding to each candidate monitoring video index information is sequentially read, and the video feature information of the monitoring video is extracted by using a pre-configured feature algorithm, according to the video feature information of the monitoring video, and Determining, by the picture feature information and the attribute information, a matching degree between the candidate monitoring video index information and the feature information, and determining whether the matching degree is greater than a first preset threshold, and if yes, selecting the candidate monitoring video The index information is used as the target monitoring video index information.
12. The method according to claim 11, wherein in the candidate monitoring video index information, the monitoring video corresponding to each candidate monitoring video index information is sequentially read, and the monitoring video is extracted by using a pre-configured feature algorithm. The video feature information, according to the video feature information of the monitoring video, and the image feature information and the attribute information, calculating a matching degree between the candidate monitoring video index information and the feature information, and determining whether the matching degree is greater than The first preset threshold, if yes, the step of using the candidate monitoring video index information as the target monitoring video index information includes:

    Constructing an initial index information set including each candidate monitoring video index information, and constructing an empty result index information set;

    Determining whether the initial index information set is empty;

    If not, the monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the initial index information set, and the video feature information of the monitoring video is extracted by using a pre-configured feature algorithm, and according to the monitoring video The video feature information, and the image feature information and the attribute information, calculate a matching degree between the candidate monitoring video index information and the feature information, and determine whether the matching degree is greater than a first preset threshold, and if yes, Adding the candidate monitoring video index information to the result index information set, deleting the candidate monitoring video index information from the initial index information set, and returning to perform the step of determining whether the initial index information set is empty;

    If so, the index information included in the result index information set is used as the target monitoring video index information.
13. The method according to claim 12, wherein a plurality of feature algorithms are pre-configured in the target storage node, and each feature algorithm has a corresponding priority, and the monitoring video is extracted by using a pre-configured feature algorithm. The steps of the video feature information include:

    Extracting video feature information of the surveillance video by using a feature algorithm with the highest priority;

    The step of using the index information included in the result index information set as the target monitoring video index information includes:

    Determining whether the result index information set is empty; when the result index information set is empty, the target storage node again constructs an initial index information set including each candidate monitoring video index information, and determines the newly constructed initial index information set. Whether it is empty;

    When the newly constructed initial index information set is not empty, the monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the newly constructed initial index information set, and the next priority feature algorithm is adopted. Extracting video feature information of the monitoring video, and calculating a matching degree between the candidate monitoring video index information and the feature information according to the video feature information of the monitoring video, the image feature information, and the attribute information, and determining If the matching degree is greater than the first preset threshold, if yes, the candidate monitoring video index information is added to the result index information set, and the candidate monitoring video index information is deleted from the newly constructed initial index information set. And returning to the step of performing the determining whether the newly constructed initial index information set is empty;

    When the newly constructed initial index information set is empty, returning to perform the step of determining whether the result index information set is empty;

    When the result index information set is not empty, the index information included in the result index information set is used as the target monitoring video index information.
14. A cloud storage system, comprising: a management node and a plurality of storage nodes;

    The management node is configured to receive, by the client, a video search request that includes feature information of the to-be-searched surveillance video, where the feature information includes at least one of: a reference picture, a spatial range, a time range, and attribute information; Determining a target storage node storing the to-be-searched surveillance video, and transmitting the video search request to each of the target storage nodes;

    Each of the target storage nodes is configured to receive the video search request, and according to the locally stored monitoring video, the stored monitoring video index information, and the feature information, the matching degree between the search and the feature information is greater than A target of a preset threshold monitors video index information, and sends the target monitoring video index information to the management node.
15. The cloud storage system according to claim 14, wherein when the feature information comprises: a reference picture, a spatial range, a time range, and attribute information,

    Any target storage node is specifically configured to perform feature extraction on the reference picture to obtain picture feature information corresponding to the reference picture; and search for candidate matching the spatial range and time range in the locally stored monitoring video index information Monitoring video index information; in the candidate monitoring video index information, sequentially reading the monitoring video corresponding to each candidate monitoring video index information, and extracting video feature information of the monitoring video by using a pre-configured feature algorithm, according to the monitoring video The video feature information, and the image feature information and the attribute information, calculate a matching degree between the candidate monitoring video index information and the feature information, and determine whether the matching degree is greater than a first preset threshold, and if yes, The candidate monitoring video index information is used as the target monitoring video index information.
16. A cloud storage system according to claim 15 wherein:

    Any target storage node, specifically configured to construct an initial index information set including each candidate monitoring video index information, and construct an empty result index information set; determine whether the initial index information set is empty; if not, at the initial The monitoring information corresponding to the first candidate monitoring video index information is sequentially read in the index information set, and the video feature information of the monitoring video is extracted by using a pre-configured feature algorithm, and according to the video feature information of the monitoring video, The picture feature information and the attribute information are used to calculate a matching degree between the candidate monitoring video index information and the feature information, and determine whether the matching degree is greater than a first preset threshold, and if so, the candidate monitoring video index information Adding the result index information set, deleting the candidate monitoring video index information from the initial index information set, and returning to perform the step of determining whether the initial index information set is empty; if yes, the result is Index information included in the index information set as the target Video index information.
17. The cloud storage system according to claim 16, wherein each storage node is pre-configured with a plurality of feature algorithms, and each feature algorithm has a corresponding priority.

    Any target storage node is specifically configured to extract video feature information of the monitoring video by using a feature algorithm with the highest priority; and when the index information included in the result index information set is used as the target monitoring video index information, Whether the result index information set is empty; when the result index information set is empty, the target storage node again constructs an initial index information set including each candidate monitoring video index information, and determines whether the newly constructed initial index information set is air;

    When the newly constructed initial index information set is not empty, the monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the newly constructed initial index information set, and the next priority feature algorithm is adopted. Extracting video feature information of the monitoring video, and calculating a matching degree between the candidate monitoring video index information and the feature information according to the video feature information of the monitoring video, the image feature information, and the attribute information, and determining If the matching degree is greater than the first preset threshold, if yes, the candidate monitoring video index information is added to the result index information set, and the candidate monitoring video index information is deleted from the newly constructed initial index information set. And returning to the step of performing the determining whether the newly constructed initial index information set is empty;

    When the newly constructed initial index information set is empty, returning to perform the step of determining whether the result index information set is empty;

    When the result index information set is not empty, the index information included in the result index information set is used as the target monitoring video index information.
18. A cloud storage system according to claim 16 or 17, wherein

    The management node is further configured to receive target monitoring video index information sent by each target storage node, and send the target monitoring video index information to the client.
19. A cloud storage system according to claim 18, wherein

    The management node is further configured to receive the acknowledgement information sent by the client, and adjust the first preset threshold in each target storage node according to the acknowledgement information.
20. A cloud storage system according to claim 19, wherein

    The management node is specifically configured to: when the accuracy rate included in the confirmation information is greater than a second preset threshold, reduce the first preset threshold; when the accuracy included in the confirmation information is less than or equal to When the second threshold is described, the first preset threshold is increased.
21. A cloud storage system according to any of claims 14-17, wherein

    The management node is specifically configured to determine, according to the spatial range, and the index information of the monitoring video stored in each locally stored storage node, the target storage node that stores the to-be-searched surveillance video.
22. A cloud storage system according to any of claims 14-17, wherein

    The management node is specifically configured to construct a target storage node list including each target storage node identifier; determine whether the target storage node list is empty; and when the target storage node list is not empty, the video search request is Sending to a target storage node in the target storage node list, and determining whether the transmission is successful; when the transmission is successful, removing the target storage node identifier from the target storage node list, and returning to perform the determining the target The step of storing whether the node list is empty; when the sending fails, returning to the step of performing the determining whether the storage node list is empty.
23. A video search device, which is applied to a target storage node in a cloud storage system, the cloud storage system further includes a management node, and the device includes:

    a receiving module, configured to receive, by the management node, a video search request that includes feature information of a to-be-searched surveillance video, where the video search request is that the management node determines that the target storage node stores a to-be-searched surveillance video, Sending to the target storage node, the feature information includes at least one of the following: a reference picture, a spatial range, a time range, and attribute information;

    a search module, configured to search, according to the locally stored monitoring video, the stored monitoring video index information, and the feature information, a target monitoring video index information that is greater than a first preset threshold between the feature information and the feature information, And transmitting the target monitoring video index information to the management node.
24. The apparatus according to claim 23, wherein when the feature information comprises: a reference picture, a spatial range, a time range, and attribute information, the searching module comprises:

    a feature extraction sub-module, configured to perform feature extraction on the reference picture to obtain picture feature information corresponding to the reference picture;

    a search submodule, configured to search, in the locally stored monitoring video index information, candidate monitoring video index information that matches the spatial range and time range;

    The search sub-module is configured to sequentially read the monitoring video corresponding to each candidate monitoring video index information in the candidate monitoring video index information, and extract the video feature information of the monitoring video by using a pre-configured feature algorithm, according to the monitoring video. The video feature information, and the image feature information and the attribute information, calculate a matching degree between the candidate monitoring video index information and the feature information, and determine whether the matching degree is greater than a first preset threshold, and if so, The candidate monitoring video index information is used as the target monitoring video index information.
25. The device according to claim 24, wherein the search submodule comprises:

    Constructing a subunit for constructing an initial index information set including each candidate monitoring video index information, and constructing an empty result index information set;

    a determining subunit, configured to determine whether the initial index information set is empty;

    a feature extraction subunit, configured to sequentially read, in the initial index information set, a monitoring video corresponding to a piece of candidate monitoring video index information in the initial index information set, using pre-configured features, when the determining subunit determines that the result is negative The algorithm extracts video feature information of the surveillance video;

    a search subunit, configured to calculate a matching degree between the candidate monitoring video index information and the feature information according to the video feature information of the monitoring video, and the image feature information and the attribute information, and determine whether the matching degree is If the value is greater than the first preset threshold, if yes, the candidate monitoring video index information is added to the result index information set, the candidate monitoring video index information is deleted from the initial index information set, and the determining subunit is triggered. ;

    Determining a subunit, configured to use index information included in the result index information set as the target monitoring video index information when the judgment subunit determines that the result is YES.
26. The device according to claim 25, wherein the feature extraction subunit is specifically configured to extract video feature information of the surveillance video by using a feature algorithm with the highest priority;

    The determining subunit is specifically configured to:

    Determining whether the result index information set is empty; when the result index information set is empty, the target storage node again constructs an initial index information set including each candidate monitoring video index information, and determines the newly constructed initial index information set. Whether it is empty;

    When the newly constructed initial index information set is not empty, the monitoring video corresponding to the first candidate monitoring video index information is sequentially read in the newly constructed initial index information set, and the next priority feature algorithm is adopted. Extracting video feature information of the monitoring video, and calculating a matching degree between the candidate monitoring video index information and the feature information according to the video feature information of the monitoring video, the image feature information, and the attribute information, and determining If the matching degree is greater than the first preset threshold, if yes, the candidate monitoring video index information is added to the result index information set, and the candidate monitoring video index information is deleted from the newly constructed initial index information set. And returning to the step of performing the determining whether the newly constructed initial index information set is empty;

    When the newly constructed initial index information set is empty, returning to perform the step of determining whether the result index information set is empty;

    When the result index information set is not empty, the index information included in the result index information set is used as the target monitoring video index information.
27. A computer device, comprising: a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory complete communication with each other through the communication bus;

    The memory is configured to store a computer program;

    The processor, when used to execute a program stored on the memory, implements the method steps of any one of claims 10-13.
28. A computer readable storage medium, for storing executable program code for performing, at runtime, according to any of claims 10-13 Method steps.
29. An application, characterized in that the application is operative to perform the method steps of any one of claims 10-13 at runtime.

Images