WO2016201992A1 - 云存储服务器的视频存储及检索方法、视频云存储系统 - Google Patents

云存储服务器的视频存储及检索方法、视频云存储系统 Download PDF

Info

Publication number
WO2016201992A1
WO2016201992A1 PCT/CN2016/072808 CN2016072808W WO2016201992A1 WO 2016201992 A1 WO2016201992 A1 WO 2016201992A1 CN 2016072808 W CN2016072808 W CN 2016072808W WO 2016201992 A1 WO2016201992 A1 WO 2016201992A1
Authority
WO
WIPO (PCT)
Prior art keywords
model data
retrieval
cloud storage
picture
sample
Prior art date
Application number
PCT/CN2016/072808
Other languages
English (en)
French (fr)
Inventor
刘莎
林起芊
王伟
闫春
汪渭春
Original Assignee
杭州海康威视数字技术股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 杭州海康威视数字技术股份有限公司 filed Critical 杭州海康威视数字技术股份有限公司
Publication of WO2016201992A1 publication Critical patent/WO2016201992A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G06F16/2453—Query optimisation
    • G06F16/24532—Query optimisation of parallel queries
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
    • G06F16/73—Querying

Definitions

  • the present application relates to a video information technology, and in particular, to a video storage and retrieval method of a cloud storage server, and a video cloud storage system.
  • Video retrieval technology has been used to improve video viewing efficiency, but usually only semantic retrieval. Many features are difficult to describe by semantics. If you can directly search according to the pictures given by the user, and avoid semantic description, it will be the most intuitive retrieval method. This program describes this retrieval method and quickly retrieves the user's needs. Data to achieve efficient mapping.
  • the video is imported when the search is needed, and the video is analyzed frame by frame, all the moving targets in the video are taken out, and recorded in a structured manner. And then store this structured description of the information for retrieval. After the sample image and the original video are structurally analyzed, the information of the sample image is used as an input to match the structured information of the original video to find a target that meets the condition.
  • the frame-by-frame analysis of the video takes a long time, either the user needs to wait for a long time to receive the processing result, thereby reducing user friendliness, or the video storage server is required to have very high processing capability. This increases the hardware investment.
  • One of the technical problems to be solved by the present application is that it is necessary to provide related technologies such as video retrieval to overcome the shortcomings of current video retrieval technology, such as low efficiency, poor user friendliness, and high equipment cost.
  • the present application first provides a video storage and retrieval method for a cloud storage server, comprising: generating a picture model based on the received original picture stream data when receiving an indication to store original picture stream data. Data; storing the original picture stream data and the picture model data; receiving a retrieval request including a sample picture; modeling the sample picture to generate sample model data; using the sample model data pair to store the stored The picture model data is retrieved to obtain a search result; the search result is sent in response to the search request.
  • searching, by using the sample model data, the stored image model data to obtain a retrieval result comprising: performing, by using the sample model data, the stored image model data that meets a retrieval condition. Retrieving, obtaining the search result; wherein the search request includes the search condition.
  • the retrieval condition comprises encoder information.
  • performing the searching on the stored image model data that meets the search condition by using the sample model data, and acquiring the search result including: storing the stored image model that meets the search condition a picture whose data is similar to the sample model data as the search result; or a video based on the stored picture model data conforming to the search condition and the sample model data and a time stamp of the pictures
  • the segment uses the video segment as the search result.
  • the retrieval condition further includes a time range.
  • the present application further provides a video cloud storage system, including multiple cloud storage servers, each cloud storage server includes a processing unit and a storage unit, wherein the processing unit is configured to receive an indication that the original image stream data is to be stored. Generating image model data based on the received original image stream data; when receiving the retrieval request including the sample image, modeling the sample image to generate sample model data; the storage unit is configured to store the The original picture stream data and the picture model data; wherein the processing unit is further configured to retrieve the stored picture model data by using the sample model data, obtain a retrieval result, and send the retrieval result to respond The retrieval request.
  • the system further includes: a cloud storage management server, configured to determine a target cloud storage server that stores the image model data that meets the retrieval condition; wherein the processing unit in the target cloud storage server utilizes the The sample model data performs the retrieval on the picture model data stored in the storage unit in the target cloud storage server that meets the retrieval condition to acquire the retrieval result; the retrieval request includes the retrieval condition.
  • a cloud storage management server configured to determine a target cloud storage server that stores the image model data that meets the retrieval condition
  • the processing unit in the target cloud storage server utilizes the The sample model data performs the retrieval on the picture model data stored in the storage unit in the target cloud storage server that meets the retrieval condition to acquire the retrieval result
  • the retrieval request includes the retrieval condition.
  • the processing unit in the target cloud storage server is configured to store the image model data that meets the retrieval condition and the sample stored by the storage unit in the target cloud storage server a picture similar to the model data as the search result; or a picture similar to the sample model data stored in the storage unit in the target cloud storage server and conforming to the search condition and the sample model data And the timestamp of the pictures, generating a video segment, and using the video segment as the search result.
  • each cloud storage server is further configured with a picture modeling switch for enabling or disabling the cloud storage server, wherein the cloud storage management server is configured to open the cloud based on the load distribution policy and the picture modeling switch.
  • a processing unit in the storage server sends the indication to store raw picture stream data.
  • the system further includes a client, wherein the client sends the retrieval request to the cloud storage management server, and receives the retrieval result; and the client, each cloud storage server and The cloud storage management server is disposed in the same or different geographical area or network area.
  • the application also provides a storage medium for saving the video of the cloud storage server mentioned above.
  • the program code executed by the storage and retrieval method.
  • the application also provides a computer terminal for executing the program code of the steps provided by the video storage and retrieval method of the cloud storage server described above.
  • the present application further provides a cloud storage server, including one or more processors, a memory, and a transmission device; wherein the memory is configured to save program code executed by the video storage and retrieval method of the cloud storage server;
  • the transmitting device is configured to receive or send data via a network;
  • the processor is configured to invoke, by the transmitting device, the information and an application stored by the memory to perform the following steps: receiving an original to be stored Generating image model data based on the received original image stream data; storing the original picture stream data and the picture model data; receiving a search request including a sample picture; constructing the sample picture And generating sample model data; searching the stored image model data by using the sample model data to obtain a retrieval result; and transmitting the retrieval result in response to the retrieval request.
  • the processor is configured to perform the following steps: searching, by using the sample model data, the stored image model data to obtain a retrieval result: using the sample model data to store the stored content that meets the retrieval condition The image model data performs the retrieval to obtain the retrieval result; wherein the retrieval request includes the retrieval condition.
  • the processor is configured to perform the following step by using the sample model data to perform the retrieval on the stored image model data that meets a search condition, and obtain the search result: the stored match is consistent with the search a picture in which the picture model data of the condition is similar to the sample model data as the retrieval result; or based on the stored picture model data conforming to the retrieval condition and the sample model data and the like The timestamp of the picture generates a video segment that is used as the search result.
  • the conventional technique of importing a post-analysis of a video when retrieval is required reduces the pressure of network bandwidth at the time of retrieval, especially the network bandwidth caused by the migration of the off-site video data.
  • the huge data processing amount caused by the need to extract the active target from the video frame by frame as the model data is required when the amount of data to be retrieved is large, and the retrieval request to the user is reduced. Time delay.
  • the solution makes full use of the cloud computing capability, utilizes the high concurrency and high reliability of the cluster, and simultaneously writes video recordings by multiple servers, analyzes the video data in real time, generates a picture stream, generates model data for the picture, and streams data with the original picture. Store together.
  • the search is needed, only the sample picture is given, and the high concurrency of the solution of the present application can be utilized to realize simultaneous work of multiple cloud storage servers, concurrently searching, and efficiently obtaining retrieval results.
  • FIG. 1 is a schematic structural diagram of a video cloud storage system according to an embodiment of the present application.
  • FIG. 2 is a schematic flowchart of generating and storing picture model data when a cloud storage server receives an indication to store original picture stream data according to an embodiment of the present application;
  • FIG. 3 is a schematic flowchart of a video storage and retrieval method of a cloud storage server according to an embodiment of the present application
  • FIG. 4 is a flow diagram of modeling sample data based on sample pictures and returning search results based on sample model data according to an embodiment of the present application.
  • the clustering mode of the cloud storage system can distribute the pressure of analyzing data and writing data to different servers, and multiple concurrent data storage and storage. Compared with the work of a single server, the efficiency is higher, and the performance is higher for the whole system. better.
  • the invention is very reasonable and efficient to set the processing body and timing of each processing step involved in video retrieval, thereby fully utilizing the concurrent processing performance advantages of the cloud storage system cluster mode. Compared with traditional video retrieval, especially for off-site retrieval can be very good and convenient.
  • FIG. 1 is a schematic structural diagram of a video cloud storage system according to an embodiment of the present application.
  • a video cloud storage system includes a cloud storage server 10 , a cloud storage management server 20 , and a client 30 .
  • the cloud storage management server 20 is connected to the client 30, and provides a write picture interface to the client 30. Based on the load distribution policy, each picture in the picture stream generated based on the video to be stored and its corresponding picture model data are displayed as pictures. Stored in units of at least one of the aforementioned one or more cloud storage servers 10.
  • the cloud storage management server 20 may be one or more (refer to VM1, VM2, ... VM2t-1 of FIG. 2, where t is a natural number).
  • the cloud storage server 10 generates picture model data based on the picture stream generated by the video to be stored, according to an instruction from the storage management server 20, and stores the generated picture model data, picture stream, and the video. Further, upon receiving the retrieval request including the sample picture, the cloud storage server 10 models the sample picture to generate sample model data, by comparing the sample model data with the stored picture model data (ie, using the sample The model data retrieves the stored picture model data to obtain a search result and returns the search result in response to the search request. Similar to the cloud storage management server 20, each cloud storage server 10 can be embodied as a physical server (refer to VS1, VS2, ... VSn of FIG. 2, where n is a natural number).
  • the cloud storage server 10 may further include a storage unit 11 and a processing unit 12.
  • Storage unit 11 is used for storing original picture stream data and picture model data generated based on the original picture stream data.
  • the processing unit 12 is configured to perform processing such as picture model data and sample model data (which will be described in detail later).
  • the client 30, each cloud storage server 10, and cloud storage management server 20 can be placed in the same or different geographic areas or network areas.
  • the user does not feel the cross area. Moreover, the retrieved results are more complete, and all data that meets the search criteria is fed back to the user. Compared with the conventional technology, when it comes to the need to search across different regions, it may need to go to different regions to view separately, or require the data of the associated regions to be collected together and then retrieved, which obviously saves manpower and Increased user friendliness.
  • the cloud storage server 10 when receiving the indication that the original picture stream data is to be stored, the cloud storage server 10 generates the picture model data based on the picture stream. Compared with the conventional technique of regenerating the picture model data of the video when receiving the user's retrieval request, the retrieval efficiency is greatly improved, and the waiting time from the time the user issues the retrieval request to the acquisition response is reduced.
  • FIG. 2 is a flow diagram of the cloud storage server 10 generating and storing picture model data upon receiving an indication to store original picture stream data.
  • FIG. 3 is a schematic flowchart diagram of a video storage and retrieval method of a cloud storage server according to an embodiment of the present application. 4 is a schematic flow chart of modeling sample data based on sample images and returning search results based on sample model data.
  • step S301 the cloud storage server 10 controls based on the load distribution policy and the picture modeling switch when receiving the indication to store the original picture stream data, so that the picture is Part or all of the cloud storage server with the modeling switch turned on generates image model data based on the received original image stream data.
  • the picture model data is generated based on the received original picture stream data upon receiving an indication to store the original picture stream data.
  • a cloud storage node (refer to VS1, VS2, ... VSn of FIG. 2) as a cloud storage server 10 is provided with a picture modeling switch, and these cloud storage nodes (at least one of The picture modeling switch is on. Or in the cloud storage system management node (refer to Vm1, Vm2, ..., Vm2-1 of FIG. 2, where m is a natural number) as the storage management server 20, respectively, a picture modeling switch for each cloud storage server 10 is provided. (For example, it can be a string of binary numbers, each bit of a binary number corresponding to a management node).
  • the cloud storage nodes (refer to VS1, VS2, ... VSn of FIG. 2) of the cloud storage servers 10, which image storage switches are turned on, Which is closed.
  • the front end of the video cloud storage system supports access to multiple types of cloud storage server devices.
  • writing the image data of the original image stream at the same time may be unnecessary for some occasions, which may be a waste of performance in some occasions. Or a large waste of storage space.
  • the video cloud storage system provides a free configuration mode, and the image modeling function can be flexibly switched through the picture modeling switch of the cloud storage server, and the granularity is refined to each cloud storage server.
  • the image storage switch of the cloud storage server can be a software switch or a hardware button.
  • the system After the image stream reaches the cloud storage system, the system allocates the image to different storage resources in a single image unit through the cluster load balancing and decentralization strategy. It can also allocate as few storage resources as possible through centralized storage.
  • Upper, decentralized or centralized storage is configurable.
  • the cloud storage server that receives the image analyzes and processes the image, and finally stores the model data and the original image stream data to realize real-time storage of the model data.
  • the picture modeling switch may not be set, so that all cloud storage servers perform the processing of generating picture model data by the picture stream.
  • Step S302 storing original picture stream data (and/or storing picture stream generated based on video data in video form) and sample model data generated in step S301.
  • the video cloud storage system When the user wants to use the picture to retrieve the video data stored by the cloud storage server 10, the video cloud storage system according to the embodiment is prompted to perform step S303.
  • Step S303 upon receiving the retrieval request including the sample picture and the retrieval condition, the cloud storage server 10 performs analysis control.
  • the sample picture is modeled upon receipt of a retrieval request including a sample picture and a retrieval condition to generate sample model data. It is analyzed which cloud storage nodes store the image stream that meets the retrieval condition, and then controls according to the analysis result, so that only the cloud storage server 10 that matches the image model data of the retrieval condition is stored in all the cloud storage servers 10 of the system.
  • the sample picture is modeled to generate sample model data.
  • the retrieval condition may include at least one of a time range and encoder information.
  • the search request may include a sample picture K, and optionally, may include a search condition such as an encoder ID and a time range.
  • search condition such as an encoder ID and a time range.
  • the cloud storage management server 20 Upon receiving the sample picture K, the cloud storage management server 20 finds all the cloud storage servers 10 storing the encoder based on the retrieval conditions such as the encoder ID and the time range. Assuming that there are a total of 50 cloud storage servers 10, each of which has 300 storage blocks, and the video data that is analyzed to obtain the retrieval conditions are respectively stored on three cloud storage servers (not shown) of A, B, and C. On the three cloud storage servers A, B, and C, there are 3, 4, and 5 blocks of video data that meet the search criteria. In this example, the cloud storage management server 20 sends the sample picture K to the three cloud storage servers A, B, and C. After receiving the sample images, the three cloud storage servers A, B, and C respectively model the sample image K to generate model data and place it in the local memory.
  • the cloud storage management server 20 sends the sample picture K to the three cloud storage servers A, B, and C. After receiving the sample images, the three cloud storage servers A, B, and C respectively model the sample image K to generate model data
  • Step S304 the cloud storage server storing the picture model data conforming to the retrieval condition obtains the retrieval result by comparing the sample model data with the picture model data stored in the storage unit and satisfying the retrieval condition.
  • the sample model data is compared with the image model data to obtain a search result.
  • the three cloud storage servers A, B, and C will retrieve the image of the encoder that is stored locally in the retrieval time range according to the model data Km in the memory, then the three servers respectively The data is retrieved only from the memory blocks that satisfy the search condition (three blocks, four blocks, and five blocks described above), and the search range is narrowed.
  • the three cloud storage servers do not interfere with each other when searching, and each server performs its duties.
  • the server will find all similar pictures in the respective machines according to the similarity of the user requests. Finally, the three cloud storage servers will find the pictures separately. Return to the upper application interface (provided by the cloud storage management server 20).
  • the search result can be a picture stream or a video segment.
  • the cloud storage server 10 may use a picture similar to the sample model data in the stored picture stream as a retrieval result.
  • the cloud storage server 10 may also generate a video segment based on information of the picture model data in the stored picture stream and the sample model data and the timestamp of the pictures, and generate the video segment.
  • a result of the search By generating a video segment, you can improve the user's feeling that the picture list is too monotonous and discontinuous.
  • the user can set the result of the request as a video segment, and the cloud storage system will integrate the picture list into a complete continuous video segment according to information such as time stamp. It can be seen that the output of the solution can meet the diverse needs of users, enabling users to find and obtain desired information more quickly.
  • step S305 the cloud storage management server 20 summarizes the search results of the cloud storage servers each storing the picture stream conforming to the retrieval condition, and returns the summarized search results in response to the retrieval request.
  • the search results of each cloud storage server are summarized, and the summarized search results are returned.
  • the cloud storage management server 20 can provide an upper application interface for receiving retrieval results from the cloud storage server. Then, the search results received by the upper application interface are summarized and fed back to the user.
  • Another advantage of the system is that it does not affect the overall result due to a single point of failure. Assume that the cloud storage server A mentioned above runs abnormally and fails to receive the retrieval result from the cloud storage server A, but this does not affect the normal retrieval of the retrieval device from the cloud storage servers B and C, and summarizes the cloud storage servers B and C. And retrieve the results and feed them back to the user.
  • the present application can concurrently work with multiple cloud storage servers to process model data in real time, thereby realizing the function of efficiently analyzing stored pictures.
  • multiple cloud storage servers to process model data in real time, thereby realizing the function of efficiently analyzing stored pictures.
  • the system will automatically notify the servers in different regions to achieve a good user experience.
  • three cloud storage servers A, B, and C are simultaneously retrieved, and by analyzing the retrieval conditions, the retrieval of the eligible storage blocks is performed in a targeted manner, and the efficiency is much higher than downloading (storing) the video to the single.
  • the server performs comparative search, which is higher than the full traversal search, and the user can get the result feedback faster.
  • the larger the amount of data the more servers, the more obvious the performance advantages of this system.
  • A, B, and C three A, B, and C cloud storage servers can be either cross-domain (geographic/network logical) or the same (geographic/network logical) region.
  • a and B belong to one area
  • C belongs to another area.
  • the three cloud storage servers of A, B, and C will still return the results to the API, and finally the upper application interface summary provided by the cloud storage management server 20. Searching the results and feeding them back to the user, once again showing a good user experience, and doing cross-regional mapping.
  • the search condition is not necessarily included in the search request. Retrieving a request including a search condition is a preferred case.
  • the cloud storage management server 20 may directly transmit the sample image data to the respective cloud storage servers 10 without analyzing which of the cloud storage servers 10 stores the video data that meets the retrieval condition.
  • Each cloud storage server 10 models the sample picture to generate sample model data, and then obtains a search result by comparing the sample model data with picture model data stored by the storage unit that conforms to the retrieval condition.
  • the solution is a video cloud storage solution that analyzes data and writes data simultaneously.
  • the image stream to be stored in the system can be automatically analyzed, the model data is generated, and stored in the cloud storage server together with the original image stream data.
  • the source of the picture stream can be a front-end capture machine or a key picture stream that is intercepted from the video stream by the cloud storage platform.
  • the original picture stream is intercepted by a cloud storage platform located on the video cloud storage system and connected to the video cloud storage system network, and the picture stream is sent to the video cloud storage system.
  • the cloud storage management server 20 of the video cloud storage system then allocates resources and finally stores them in the cloud storage server 10.
  • the cloud storage server 10 writes the model data together with the original image stream data into the cloud storage server, thereby eliminating the need to import the original image stream data into the server for analysis and processing before the traditional method, and saves waiting for the original.
  • the time when the image stream data is modeled separately can improve the retrieval efficiency.
  • the sample image analysis given by the user is firstly analyzed, and then the sample image model data is used to directly retrieve the image satisfying the demand from the model data stored in the system, and the map is searched. Only the sample images need to be modeled in real time during the retrieval. It is not necessary to wait for all the original image stream data modeling to complete the comparison, so that the scheme is efficient and simplifies the instant processing.
  • An embodiment of the present application further provides a storage medium for saving the foregoing cloud storage server.
  • the program code executed by the video storage and retrieval method.
  • the foregoing storage medium may be configured to be configured to execute to generate a picture model based on the received original picture stream data when receiving an indication that the original picture stream data is to be stored. Data; storing the original picture stream data and the picture model data; receiving a retrieval request including a sample picture; modeling the sample picture to generate sample model data; using the sample model data pair to store the stored The picture model data is retrieved to obtain a search result; the search result is sent in response to the search request.
  • the storage medium is further configured to store program code for performing the following step by using the sample model data to perform the retrieval on the stored image model data that meets the retrieval condition, and acquiring the retrieval result; Wherein the retrieval request includes the retrieval condition.
  • the storage medium is further configured to store program code for performing the following steps: using the stored picture of the picture model data that meets the retrieval condition and the sample model data as the retrieval result;
  • a video segment is generated based on the stored pictures in which the picture model data conforming to the retrieval condition is similar to the sample model data and time stamps of the pictures, and the video segment is used as the retrieval result.
  • the embodiment of the present application further provides a computer terminal for executing the program code of the steps provided by the video storage and retrieval method of the cloud storage server.
  • the embodiment of the present application further provides that the cloud storage server includes one or more processors, a memory, and a transmission device.
  • the memory can be used to store the software program and the program block, such as the program code executed by the video storage and retrieval method of the cloud storage server in the embodiment of the present application, and the processor runs the software program and the module block stored in the memory. , thereby performing various functional applications and data processing, that is, implementing the video storage and retrieval method of the cloud storage server described above.
  • the memory may include a high speed random access memory, and may also include non-volatile memory such as one or more magnetic storage devices, flash memory, or other non-volatile solid state memory.
  • the memory can further include memory remotely located relative to the processor, which can be connected to the terminal over a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.
  • the above transmission device is for receiving or transmitting data via a network.
  • the above network Examples of the body may include a wired network and a wireless network.
  • the transmission device includes a Network Interface Controller (NIC) that can be connected to other network devices and routers via a network cable to communicate with the Internet or a local area network.
  • the transmission device is a Radio Frequency (RF) module for communicating with the Internet wirelessly.
  • NIC Network Interface Controller
  • RF Radio Frequency
  • the memory is used to store preset action conditions and information of the preset rights user, and an application.
  • the processor may invoke the information and the application stored by the memory by the transmitting device to perform the step of: generating a picture model based on the received original picture stream data upon receiving an indication that the original picture stream data is to be stored Data; storing the original picture stream data and the picture model data; receiving a retrieval request including a sample picture; modeling the sample picture to generate sample model data; using the sample model data pair to store the stored The picture model data is retrieved to obtain a search result; the search result is sent in response to the search request.
  • the processor is configured to perform the following steps: searching, by using the sample model data, the stored image model data to obtain a retrieval result: using the sample model data to store the stored content that meets the retrieval condition The image model data performs the retrieval to obtain the retrieval result; wherein the retrieval request includes the retrieval condition.
  • the processor is configured to perform the following step by using the sample model data to perform the searching on the stored image model data that meets the retrieval condition to obtain the retrieval result:

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computational Linguistics (AREA)
  • Multimedia (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Two-Way Televisions, Distribution Of Moving Picture Or The Like (AREA)

Abstract

本申请公开了一种云存储服务器的视频存储及检索方法、视频云存储系统，克服当前视频检索技术效率较低等不足。该方法包括：在接收到要存储原始图片流数据的指示时，基于所接收到的原始图片流数据生成图片模型数据；存储所述原始图片流数据和所述图片模型数据；接收包括样本图片的检索请求；对所述样本图片进行建模，生成样本模型数据；利用所述样本模型数据对所存储的所述图片模型数据进行检索，获取检索结果；发送所述检索结果以响应所述检索请求。

Description

云存储服务器的视频存储及检索方法、视频云存储系统
本申请要求于2015年6月17日提交中国专利局、申请号为201510341501.0发明名称为“云存储服务器的视频存储及检索方法、视频云存储系统”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及一种视频信息技术，尤其涉及云存储服务器的视频存储及检索方法、视频云存储系统。
背景技术
随着云技术的发展，云存储应用也逐渐渗入各个领域，数据爆炸性增长，海量的监控设备每天产生的视频数据不可估量。在公共安全领域，通过视频侦查技术查案已经是很常用的技术。面对海量的视频数据，甚至是跨区域的数据，用户如何高效的获取到想要的信息，成为一个很明显的问题。
目前已有通过视频检索技术来提高视频查看效率，但是通常只是语义检索。很多特征难以用语义进行描述，如果可以根据用户给出的图片直接进行检索，免去语义描述，将会是最直观的检索方式，本方案即讲述此种检索方式，快速检索出符合用户需求的数据，实现高效的以图搜图。
然而，在现有的通过样本图片检索图片或视频的方案中，是在需要检索时才将视频导入，并对视频逐帧进行分析，将视频中所有活动目标取出，并以结构化的方式记录，然后将此结构化描述的信息入库，供信息检索。当样本图片与原始视频都结构化分析完毕后，才以样本图片的信息作为输入，与原始视频的结构化信息进行匹配，查找符合条件的目标。该技术方案中，由于对视频逐帧分析需要占用较长的时间，因而要么导致用户需要经过漫长的等待才能够接收到处理结果从而降低用户友好性，要么要求视频存储服务器具有非常高的处理能力了从而增加硬件投入。
此外，存在一种在基于图片搜索时进行二级检索的现有技术。该现有 技术，为每个图片设置相同或不同的图片编号，通过图片编号建立图片与视频的对应关系。在检索到相似图片时，通过其图片编号来索引与之对应的视频地址，以确定作为检索结果的视频。在该技术中，虽然可以返回视频检索结果，但是，其返回的是整段视频结果，导致用户可能需要在几个小时长的视频中找到其期望看到的几秒中的视频段，这对显然给用户带来了巨大的困扰。
发明内容
本申请所要解决的技术问题之一是需要提供视频检索等相关技术，来克服当前视频检索技术效率较低、用户友好性较差以及设备成本较高等不足。
为了解决上述技术问题，本申请首先提供了一种云存储服务器的视频存储及检索方法，包括：在接收到要存储原始图片流数据的指示时，基于所接收到的原始图片流数据生成图片模型数据；存储所述原始图片流数据和所述图片模型数据；接收包括样本图片的检索请求；对所述样本图片进行建模，生成样本模型数据；利用所述样本模型数据对所存储的所述图片模型数据进行检索，获取检索结果；发送所述检索结果以响应所述检索请求。
可选地，利用所述样本模型数据对所存储的所述图片模型数据进行检索，获取检索结果，包括：利用所述样本模型数据对所存储的符合检索条件的所述图片模型数据进行所述检索，获取所述检索结果；其中，所述检索请求包括所述检索条件。
可选地，所述检索条件包括编码器信息。
可选地，利用所述样本模型数据对所存储的符合检索条件的所述图片模型数据进行所述检索，获取所述检索结果，包括：将所存储的符合所述检索条件的所述图片模型数据与所述样本模型数据相似的图片作为所述检索结果；或者，基于所存储的符合所述检索条件的所述图片模型数据与所述样本模型数据相似的图片及这些图片的时间戳生成视频段，将所述视频段作为所述检索结果。
可选地，所述检索条件进一步包括时间范围。
本申请还提供了一种视频云存储系统，包括多台云存储服务器，每台云存储服务器包括处理单元和存储单元，其中，所述处理单元用于在接收到要存储原始图片流数据的指示时，基于所接收到的原始图片流数据生成图片模型数据；在接收到包括样本图片的检索请求时，对所述样本图片进行建模，生成样本模型数据；所述存储单元用于存储所述原始图片流数据和所述图片模型数据；其中，所述处理单元还用于利用所述样本模型数据对所存储的所述图片模型数据进行检索，获取检索结果，并发送所述检索结果以响应所述检索请求。
可选地，所述系统还包括：云存储管理服务器，用于确定存储符合检索条件的图片模型数据的目标云存储服务器；其中，所述目标云存储服务器中的所述处理单元，利用所述样本模型数据对所述目标云存储服务器中的所述存储单元存储的符合所述检索条件的图片模型数据进行所述检索，获取所述检索结果；所述检索请求包括所述检索条件。
可选地，所述目标云存储服务器中的所述处理单元，用于将所述目标云存储服务器中的所述存储单元存储的、符合所述检索条件的所述图片模型数据与所述样本模型数据相似的图片，作为所述检索结果；或者，基于所述目标云存储服务器中的所述存储单元存储的、符合所述检索条件的所述图片模型数据与所述样本模型数据相似的图片、及这些图片的时间戳，生成视频段，将所述视频段作为所述检索结果。
可选地，每台云存储服务器还设置有开启或关闭云存储服务器的图片建模开关，其中，所述云存储管理服务器用于基于负载分配策略和所述图片建模开关，向开启的云存储服务器中的处理单元发送存储原始图片流数据的所述指示。
可选地，该系统还包括客户端，其中，所述客户端向所述云存储管理服务器发送所述检索请求，并接收所述检索结果；以及，所述客户端、每台云存储服务器与所述云存储管理服务器被设置在相同或不同的地理区域或网络区域。
本申请还提供了一种存储介质，用于保存上述的云存储服务器的视频 存储及检索方法所执行的程序代码。
本申请还提供了一种计算机终端，用于执行上述的云存储服务器的视频存储及检索方法提供的步骤的程序代码。
本申请还提供了一种云存储服务器，包括一个或多个处理器、存储器以及传输装置；其中，所述存储器，用于保存上述的云存储服务器的视频存储及检索方法所执行的程序代码；所述传输装置，用于经由一个网络接收或者发送数据；所述处理器，用于通过所述传输装置调用所述存储器存储的信息及应用程序，以执行下述步骤：在接收到要存储原始图片流数据的指示时，基于所接收到的原始图片流数据生成图片模型数据；存储所述原始图片流数据和所述图片模型数据；接收包括样本图片的检索请求；对所述样本图片进行建模，生成样本模型数据；利用所述样本模型数据对所存储的所述图片模型数据进行检索，获取检索结果；发送所述检索结果以响应所述检索请求。
可选地，所述处理器用于执行以下步骤利用所述样本模型数据对所存储的所述图片模型数据进行检索，获取检索结果：利用所述样本模型数据对所存储的符合检索条件的所述图片模型数据进行所述检索，获取所述检索结果；其中，所述检索请求包括所述检索条件。
可选地，所述处理器用于执行以下步骤利用所述样本模型数据对所存储的符合检索条件的所述图片模型数据进行所述检索，获取所述检索结果：将所存储的符合所述检索条件的所述图片模型数据与所述样本模型数据相似的图片作为所述检索结果；或者，基于所存储的符合所述检索条件的所述图片模型数据与所述样本模型数据相似的图片及这些图片的时间戳生成视频段，将所述视频段作为所述检索结果。
根据本申请的一方面，相比在需要检索时才导入视频的后分析的传统技术，降低了检索时的数据迁移尤其是异地视频数据迁移带来的网络带宽的压力。
此外，根据本申请的一方面，避免了在当检索数据量大时因需要即时从视频中逐帧分析取出活动目标作为模型数据所带来的突发巨大数据处理量，降低了对用户检索请求的时间延迟。
另一方面，本申请的一个可选方案中，无论是对原始图片流数据进行分析还是将样本图片与已结构化模型数据进行对比，都体现了并发性，进而能够提高分析效率和提高检索效率。
综上，本方案充分利用云计算能力，利用集群的高并发、高可靠性，多台服务器写录像的同时，实时分析视频数据，生成图片流，对图片生成模型数据，并与原始图片流数据一起存储。在需要检索时，只需给出样本图片，即可利用本申请方案的高并发性，实现多个云存储服务器同时工作，并发地检索，高效得出检索结果。
本申请的其他优点、目标和特征在某种程度上将在随后的说明书中进行阐述，并且在某种程度上，基于对下文的考察研究对本领域技术人员而言将是显而易见的，或者可以从本申请的实践中得到教导。本申请的目标和其他优点可以通过下面的说明书，权利要求书，以及附图中所特别指出的结构来实现和获得。
附图说明
附图用来提供对本申请的进一步理解，并且构成说明书的一部分，与本申请的实施例共同用于解释本申请，并不构成对本申请的限制。
图1是根据本申请实施例的视频云存储系统的架构示意图；
图2是根据本申请实施例的云存储服务器在接收到要存储原始图片流数据的指示时生成及存储图片模型数据的流程示意图；
图3是根据本申请实施例的云存储服务器的视频存储及检索方法的流程示意图；
图4是根据本申请实施例基于样本图片进行建模获取样本模型数据并基于样本模型数据返回检索结果的流程示意图。
具体实施方式
以下将结合附图及实施例来详细说明本申请的实施方式，借此对本申请如何应用技术手段来解决技术问题，并达成技术效果的实现过程能充分 理解并据以实施。需要说明的是，只要不构成冲突，本申请中的各个实施例以及各实施例中的各个特征可以相互结合，所形成的技术方案均在本申请的保护范围之内。
另外，在附图的流程图示出的步骤可以在诸如一组计算机可执行指令的计算机系统中执行，并且，虽然在流程图中示出了逻辑顺序，但是在某些情况下，可以以不同于此处的顺序执行所示出或描述的步骤。
云存储系统的集群模式能够将分析数据和写数据的压力分摊到不同的服务器上，多台并发对数据分析存储，相比单台服务器的工作，效率更高，对整个系统而言，就是性能更优。本方发明非常合理高效的设置了视频检索涉及的各个处理步骤的处理主体与时机，从而充分利用了云存储系统集群模式的并发处理性能优势。相比传统的视频检索而言，尤其针对异地检索能非常好的体现出便捷效率。
图1是根据本申请实施例的视频云存储系统的架构示意图。
如图1所示，根据本申请实施例的视频云存储系统包括云存储服务器10、云存储管理服务器20和客户端30。
云存储管理服务器20，与客户端30数据连接，向客户端30提供写图片接口，基于负载分配策略将基于要存储的视频所生成的图片流中的各个图片及其对应的图片模型数据以图片为单位，存储到前述一个或多个云存储服务器10的至少之一。云存储管理服务器20可以为一个或多个(参考图2的VM1，VM2，……VM2t-1，其中，t为自然数)。
云存储服务器10根据存储管理服务器20的指示，基于要存储的由视频生成的图片流，生成图片模型数据，存储所生成的图片模型数据、图片流和该视频。此外，在接收包括样本图片的检索请求时，云存储服务器10对所述样本图片进行建模，生成样本模型数据，通过将该样本模型数据与所存储的图片模型数据进行比较(也即利用样本模型数据对所存储的图片模型数据进行检索)，来获取检索结果，并返回该检索结果以响应于所述检索请求。和云存储管理服务器20类似，各个云存储服务器10可以具体实现为物理服务器(参考图2的VS1，VS2，……VSn，其中n为自然数)。
云存储服务器10可以进一步包括存储单元11和处理单元12。存储单元 11用于存储原始图片流数据和基于原始图片流数据生成的图片模型数据。处理单元12用于执行图片模型数据和样本模型数据(后续将详细说明)等处理。
客户端30、每台云存储服务器10和云存储管理服务器20，可被设置在相同或不同的地理区域或网络区域。
通过本申请，即便每台云存储服务器10和云存储管理服务器20被布置在不同的地理区域或网络区域，用户也感觉不到跨区域。而且检索到的结果是更完整的，所有满足检索条件的数据都会反馈给用户。这相比在传统技术中当遇到需要跨区域异地检索时可能需要到不同的区域分别去查看、或者要求将关联区域的数据全部集中在一起再检索的技术方案，显然既节约了人力、又提高了用户友好性。
如前面所述，本申请的实施例中，云存储服务器10在接收到要存储原始图片流数据的指示时，便基于所述图片流生成图片模型数据。这相比在接收到用户的检索请求时再生成视频的图片模型数据的传统技术，大大提高了检索效率，减少了用户从发出检索请求到获取响应期间的等待时长。
下面结合附图2-4来重点说明云存储服务器10在接收到要存储原始图片流数据的指示时生成图片模型数据的流程。
图2是云存储服务器10在接收到要存储原始图片流数据的指示时生成及存储图片模型数据的流程示意图。图3是根据本申请实施例的云存储服务器的视频存储及检索方法的流程示意图。图4是基于样本图片进行建模获取样本模型数据并基于样本模型数据返回检索结果的流程示意图。
在图3所示的云存储服务器的视频存储及检索方法中，步骤S301，云存储服务器10在接收到要存储原始图片流数据的指示时基于负载分配策略和图片建模开关进行控制，使得图片建模开关为开启的部分或全部云存储服务器才基于所接收到的原始图片流数据生成图片模型数据。在接收到要存储原始图片流数据的指示时基于所接收到的原始图片流数据生成图片模型数据。
具体的，参考图2，作为云存储服务器10的云存储节点(参考图2的VS1，VS2，……VSn)设置有图片建模开关，且这些云存储节点(至少之一 的图片建模开关为开启。或者在作为存储管理服务器20的云存储系统管理节点(参考图2的Vm1，Vm2，……Vm2-1，其中m为自然数)中设置有分别针对与每台云存储服务器10的图片建模开关(例如，可以为一串二进制数，二进制数的每一位对应一个管理节点)。在接收到要存储原始视频数据的指示时，获知作为云存储服务器10的各个云存储节点(参考图2的VS1，VS2，……VSn)中，哪些云存储节点的图片建模开关为开启，哪些为关闭。
在本申请中，根据本实施例的视频云存储系统前端支持接入多种类型的云存储服务器设备。
如果所有的云存储节点全部对图片开放建模功能，写入原始图片流数据的同时写入图片模型数据，对于某些场合可能是不必要的，那样在某些场合可能会是对性能的浪费或对存储空间的较大浪费。
因此，根据本实施例的视频云存储系统提供了自由配置方式，可以通过云存储服务器的图片建模开关来灵活的开关图片建模功能，粒度细化到每台云存储服务器。
云存储服务器的图片建模开关即可以为软件开关，也可以实现为硬件按钮。
当图片流到达云存储系统后，系统会通过集群负载均衡与分散策略，以单张图片为单位，将图片分配到不同的存储资源上，也可以通过集中存储方式，尽量分配到少的存储资源上，分散或集中存储是可配置的。接收到图片的云存储服务器会对图片进行分析建模处理，最后一并存储模型数据与原始图片流数据，实现模型数据的实时存储。
当然，在有些场合，也可以不设置图片建模开关，使得所有云存储服务器均会执行图片流生成图片模型数据的处理。
步骤S302，存储原始图片流数据(和/或存储基于视频形式的视频数据生成的图片流)和步骤S301中生成的样本模型数据。
在用户要利用图片来检索云存储服务器10存储的视频数据时，则促发根据本实施例的视频云存储系统执行步骤S303。
步骤S303，在接收到包括样本图片和检索条件的检索请求时，云存储服务器10进行分析控制。在接收到包括样本图片和检索条件的检索请求时对所述样本图片进行建模以生成样本模型数据。分析哪些云存储节点中存储有符合检索条件的图片流，然后根据分析结果进行控制，使得在系统的所有云存储服务器10中仅仅存储有符合所述检索条件的图片模型数据的云存储服务器10才对所述样本图片进行建模，以生成样本模型数据。其中，检索条件可包括时间范围和编码器信息至少之一。
例如，当用户有需求要做图片检索时，发出检索请求。该检索请求中可包括样本图片K，可选的，还可包括编码器ID、时间范围等检索条件。在指定时间范围时，可通过缩小时间范围来提高检索效率。
收到样本图片K后，云存储管理服务器20根据编码器ID、时间范围等检索条件查找到所有存储有该编码器的云存储服务器10。假设总共有50台云存储服务器10，每台服务器分别有300个存储块，分析得到满足检索条件的视频数据分别存储在A、B、C三台云存储服务器(未示出)上。而在A、B、C三台云存储服务器上，分别有3块、4块、5块存在符合检索条件的视频数据。在这个实例中，云存储管理服务器20将样本图片K发送到A、B、C三台云存储服务器。A、B、C三台云存储服务器接收到样本图片后，分别对样本图片K进行建模，生成模型数据，放于本地内存中。
步骤S304，存储有符合所述检索条件的图片模型数据的云存储服务器通过将样本模型数据与其存储单元存储的符合所述检索条件的图片模型数据进行比较来获取检索结果。将样本模型数据与图片模型数据进行比较来获取检索结果。
针对上面的例子，在本步骤中，A、B、C三台云存储服务器会根据内存中的模型数据Km，并行检索存储在本地满足检索时间范围内该编码器的图片，那么三台服务器分别只从符合所述检索条件的存储块中取出数据来检索(分别为上述的3块、4块、5块)，缩小检索范围。三台云存储服务器检索时互不干扰，各司其职，服务器会根据用户请求的相似度，找出各自机器中所有满足条件的相似图片，最终三台云存储服务器会分别将找出的图片返回给上层应用程序接口(可由云存储管理服务器20提供)。
检索结果可以为图片流，也可以为视频段。云存储服务器10可将所存储的图片流中图片模型数据与所述样本模型数据相似的图片作为检索结果。在另一种方案中，云存储服务器10也可将基于所存储的图片流中图片模型数据与样本模型数据相似的图片及这些图片的时间戳等信息来生成视频段，将所生成的视频段作为检索结果。通过生成视频段，可以改善图片列表过于单调、不连续的用户感受。用户可以将请求的结果设置为视频段，那么云存储系统会将图片列表根据时间戳等信息整合为完整连续的视频段。可见，本方案的输出结果可以满足用户多样化的需求，使用户能多种手段更快地查找并获得想要的信息。
步骤S305，云存储管理服务器20汇总各个存储有符合检索条件的图片流的云存储服务器的检索结果，并响应于检索请求返回汇总后的检索结果。汇总各个云存储服务器的检索结果，并返回汇总后的检索结果。例如，云存储管理服务器20可以提供用于接收到来自云存储服务器的检索结果的上层应用程序接口。然后对上层应用程序接口接收的检索结果进行汇总，反馈给用户。
此外，本系统的又一优点为，不会因为单点故障影响整体结果。假设上述提及的云存储服务器A运行异常，未能接收到来自云存储服务器A的检索结果，但这并不影响从云存储服务器B和C正常接收检索设备，并汇总云存储服务器B、C及检索结果并将之反馈给用户。
根据上述步骤S303到S304可知，本申请可以多台云存储服务器并发工作，实时处理模型数据，实现了高效分析存储图片的功能。此外，对于跨区域异地检索，对本方案而言与本地检索在用户体验上是无差异的，系统会自动通知到不同区域的服务器，做到良好的用户体验。
在上面的例子中，A、B和C三台云存储服务器同时检索，而且通过分析检索条件，针对性地对符合条件的存储块进行检索，效率远远高于将视频下载(存储)到单台服务器来进行比较检索，高于全遍历检索，用户能够更快得到结果反馈。数据量越大，服务器越多，本系统的性能优势越明显。
在上面的例子中，A、B、C三台A、B、C三台云存储服务器既可以是跨(地理/网络逻辑)区域的，也可以是同一(地理/网络逻辑)区域的。对于跨区 域异地检索，比如A、B属于一个区域，C属于另一个区域，A、B、C三台云存储服务器还是会正常返回结果给API，最终由云存储管理服务器20提供的上层应用程序接口汇总检索结果并将之反馈给用户，再次表现了良好的用户体验，做到了跨区域的以图搜图。
需要说明的是，不要求检索请求中一定包括检索条件。检索请求包括检索条件是一个优选情况。当检索请求不包括检索条件时，云存储管理服务器20可在不分析哪些云存储服务器10存储了符合检索条件的视频数据的情况下，直接将样本图片数据发送给各个云存储服务器10。各个云存储服务器10对所述样本图片进行建模，以生成样本模型数据，然后通过将所述样本模型数据与其存储单元存储的符合所述检索条件的图片模型数据进行比较来获取检索结果。
综上可知，本方案是一种分析数据与写数据同时进行的视频云存储方案。根据本申请实施例视频云存储系统的技术方案，能够自动对要存储至该系统中的图片流进行智能分析、生成模型数据，并与原始图片流数据一并存储到云存储服务器中。图片流的来源可以是前端的抓拍机或者是通过云存储平台从视频流中截取出来的关键图片流。由位于视频云存储系统之上且与视频云存储系统网络连接的云存储平台截取原始图片流，并将图片流发送到视频云存储系统。然后经过视频云存储系统的云存储管理服务器20分配资源，最后存储到云存储服务器10中。云存储服务器10将这种模型数据与原始图片流数据一并写入云存储服务器的方式，免去了传统方式检索前需要将原始图片流数据再次导入服务器去分析处理的过程，节约了等待原始图片流数据单独建模的时间，能够提高检索效率。
当使用本方案执行图片检索时，会首先对用户给出的样本图片分析建模，再利用样本图片模型数据直接从系统之前存储的模型数据中检索出满足需求的图片，实现以图搜图。检索时只有样本图片需要即时建模，无需像传统的方式需要等所有的原始图片流数据建模完成，才能去比对检索，这样的方案既高效又能简化即时处理流程。
本申请的实施例还提供了一种存储介质，用于保存上述云存储服务器的 视频存储及检索方法所执行的程序代码。可选地，在本实施例中，上述存储介质可以被设置为存储用于执行以下程序代码：在接收到要存储原始图片流数据的指示时，基于所接收到的原始图片流数据生成图片模型数据；存储所述原始图片流数据和所述图片模型数据；接收包括样本图片的检索请求；对所述样本图片进行建模，生成样本模型数据；利用所述样本模型数据对所存储的所述图片模型数据进行检索，获取检索结果；发送所述检索结果以响应所述检索请求。
可选地，存储介质还被设置为存储用于执行以下步骤的程序代码：利用所述样本模型数据对所存储的符合检索条件的所述图片模型数据进行所述检索，获取所述检索结果；其中，所述检索请求包括所述检索条件。
可选地，存储介质还被设置为存储用于执行以下步骤的程序代码：将所存储的符合所述检索条件的所述图片模型数据与所述样本模型数据相似的图片作为所述检索结果；或者，基于所存储的符合所述检索条件的所述图片模型数据与所述样本模型数据相似的图片及这些图片的时间戳生成视频段，将所述视频段作为所述检索结果。
本申请的实施例还提供了一种计算机终端，用于执行上述云存储服务器的视频存储及检索方法提供的步骤的程序代码。
可选地，本申请的实施例还提供了云存储服务器包括一个或多个处理器、存储器以及传输装置。
其中，存储器可用于存储软件程序以及模程序块，如本申请实施例中的云存储服务器的视频存储及检索方法所执行的程序代码，处理器通过运行存储在存储器内的软件程序以及模程序块，从而执行各种功能应用以及数据处理，即实现上述的云存储服务器的视频存储及检索方法。存储器可包括高速随机存储器，还可以包括非易失性存储器，如一个或者多个磁性存储装置、闪存、或者其他非易失性固态存储器。在一些实例中，存储器可进一步包括相对于处理器远程设置的存储器，这些远程存储器可以通过网络连接至终端。上述网络的实例包括但不限于互联网、企业内部网、局域网、移动通信网及其组合。
上述的传输装置用于经由一个网络接收或者发送数据。上述的网络具 体实例可包括有线网络及无线网络。在一个实例中，传输装置包括一个网络适配器(Network Interface Controller，NIC)，其可通过网线与其他网络设备与路由器相连从而可与互联网或局域网进行通讯。在一个实例中，传输装置为射频(Radio Frequency，RF)模块，其用于通过无线方式与互联网进行通讯。
其中，具体地，存储器用于存储预设动作条件和预设权限用户的信息、以及应用程序。
处理器可以通过所述传输装置调用所述存储器存储的信息及应用程序，以执行下述步骤：在接收到要存储原始图片流数据的指示时，基于所接收到的原始图片流数据生成图片模型数据；存储所述原始图片流数据和所述图片模型数据；接收包括样本图片的检索请求；对所述样本图片进行建模，生成样本模型数据；利用所述样本模型数据对所存储的所述图片模型数据进行检索，获取检索结果；发送所述检索结果以响应所述检索请求。
可选地，所述处理器用于执行以下步骤利用所述样本模型数据对所存储的所述图片模型数据进行检索，获取检索结果：利用所述样本模型数据对所存储的符合检索条件的所述图片模型数据进行所述检索，获取所述检索结果；其中，所述检索请求包括所述检索条件。
可选地，所述处理器用于执行以下步骤利用所述样本模型数据对所存储的符合检索条件的所述图片模型数据进行所述检索，获取所述检索结果：
将所存储的符合所述检索条件的所述图片模型数据与所述样本模型数据相似的图片作为所述检索结果；或者，基于所存储的符合所述检索条件的所述图片模型数据与所述样本模型数据相似的图片及这些图片的时间戳生成视频段，将所述视频段作为所述检索结果。
虽然本申请所揭露的实施方式如上，但所述的内容只是为了便于理解本申请而采用的实施方式，并非用以限定本申请。任何本申请所属技术领域内的技术人员，在不脱离本申请所揭露的精神和范围的前提下，可以在实施的形式上及细节上作任何的修改与变化，但本申请的专利保护范围，仍须以所附的权利要求书所界定的范围为准。

Claims (15)

  1. 一种云存储服务器的视频存储及检索方法，其特征在于，包括：
    在接收到要存储原始图片流数据的指示时，基于所接收到的原始图片流数据生成图片模型数据；
    存储所述原始图片流数据和所述图片模型数据；
    接收包括样本图片的检索请求；
    对所述样本图片进行建模，生成样本模型数据；
    利用所述样本模型数据对所存储的所述图片模型数据进行检索，获取检索结果；
    发送所述检索结果以响应所述检索请求。
  2. 根据权利要求1所述的方法，其特征在于，利用所述样本模型数据对所存储的所述图片模型数据进行检索，获取检索结果，包括：
    利用所述样本模型数据对所存储的符合检索条件的所述图片模型数据进行所述检索，获取所述检索结果；
    其中，所述检索请求包括所述检索条件。
  3. 根据权利要求2所述的方法，其特征在于，所述检索条件包括编码器信息。
  4. 根据权利要求2所述的方法，其特征在于，利用所述样本模型数据对所存储的符合检索条件的所述图片模型数据进行所述检索，获取所述检索结果，包括：
    将所存储的符合所述检索条件的所述图片模型数据与所述样本模型数据相似的图片作为所述检索结果；或者，
    基于所存储的符合所述检索条件的所述图片模型数据与所述样本模型数据相似的图片及这些图片的时间戳生成视频段，将所述视频段作为所述检索结果。
  5. 根据权利要求2、3或4所述的方法，其特征在于，所述检索条件进一步包括时间范围。
  6. 一种视频云存储系统，其特征在于，包括多台云存储服务器，每台云存储服务器包括处理单元和存储单元，其中，
    所述处理单元用于在接收到要存储原始图片流数据的指示时，基于所接收到的原始图片流数据生成图片模型数据；在接收到包括样本图片的检索请求时，对所述样本图片进行建模，生成样本模型数据；
    所述存储单元用于存储所述原始图片流数据和所述图片模型数据；
    其中，所述处理单元还用于利用所述样本模型数据对所存储的所述图片模型数据进行检索，获取检索结果，并发送所述检索结果以响应所述检索请求。
  7. 根据权利要求6所述的系统，其特征在于，所述系统还包括：
    云存储管理服务器，用于确定存储符合检索条件的图片模型数据的目标云存储服务器；
    其中，所述目标云存储服务器中的所述处理单元，利用所述样本模型数据对所述目标云存储服务器中的所述存储单元存储的符合所述检索条件的图片模型数据进行所述检索，获取所述检索结果；所述检索请求包括所述检索条件。
  8. 根据权利要求7所述的系统，其特征在于：
    所述目标云存储服务器中的所述处理单元，用于将所述目标云存储服务器中的所述存储单元存储的、符合所述检索条件的所述图片模型数据与所述样本模型数据相似的图片，作为所述检索结果；或者，
    基于所述目标云存储服务器中的所述存储单元存储的、符合所述检索条件的所述图片模型数据与所述样本模型数据相似的图片、及这些图片的时间戳，生成视频段，将所述视频段作为所述检索结果。
  9. 根据权利要求7所述的系统，其特征在于，每台云存储服务器还设置有开启或关闭云存储服务器的图片建模开关，其中，
    所述云存储管理服务器基于负载分配策略和所述图片建模开关，向开启的云存储服务器中的处理单元发送存储原始图片流数据的所述指示。
  10. 根据权利要求9所述的系统，其特征在于，还包括客户端，其中，
    所述客户端向所述云存储管理服务器发送所述检索请求，并接收所述检索结果；以及，
    所述客户端、每台云存储服务器与所述云存储管理服务器被设置在相同或不同的地理区域或网络区域。
  11. 一种存储介质，其特征在于，用于保存所述权利要求1所述的云存储服务器的视频存储及检索方法所执行的程序代码。
  12. 一种计算机终端，其特征在于，用于执行所述权利要求1所述的云存储服务器的视频存储及检索方法提供的步骤的程序代码。
  13. 一种云存储服务器，其特征在于，包括一个或多个处理器、存储器以及传输装置；其中，
    所述存储器，用于保存所述权利要求1所述的云存储服务器的视频存储及检索方法所执行的程序代码；
    所述传输装置，用于经由一个网络接收或者发送数据；
    所述处理器，用于通过所述传输装置调用所述存储器存储的信息及应用程序，以执行下述步骤：在接收到要存储原始图片流数据的指示时，基于所接收到的原始图片流数据生成图片模型数据；存储所述原始图片流数据和所述图片模型数据；接收包括样本图片的检索请求；对所述样本图片进行建模，生成样本模型数据；利用所述样本模型数据对所存储的所述图片模型数据进行检索，获取检索结果；发送所述检索结果以响应所述检索请求。
  14. 根据权利要求13所述的云存储服务器，其特征在于，所述处理器用于执行以下步骤利用所述样本模型数据对所存储的所述图片模型数据进行检索，获取检索结果：
    利用所述样本模型数据对所存储的符合检索条件的所述图片模型数据进行所述检索，获取所述检索结果；
    其中，所述检索请求包括所述检索条件。
  15. 根据权利要求14所述的云存储服务器，其特征在于，所述处理器用于执行以下步骤利用所述样本模型数据对所存储的符合检索条件的所述图片模型数据进行所述检索，获取所述检索结果：
    将所存储的符合所述检索条件的所述图片模型数据与所述样本模型数据相似的图片作为所述检索结果；或者，
    基于所存储的符合所述检索条件的所述图片模型数据与所述样本模型数据相似的图片及这些图片的时间戳生成视频段，将所述视频段作为所述检索结果。
PCT/CN2016/072808 2015-06-17 2016-01-29 云存储服务器的视频存储及检索方法、视频云存储系统 WO2016201992A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201510341501.0A CN106257447B (zh) 2015-06-17 2015-06-17 云存储服务器的视频存储及检索方法、视频云存储系统
CN201510341501.0 2015-06-17

Publications (1)

Publication Number Publication Date
WO2016201992A1 true WO2016201992A1 (zh) 2016-12-22

Family

ID=57544899

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/072808 WO2016201992A1 (zh) 2015-06-17 2016-01-29 云存储服务器的视频存储及检索方法、视频云存储系统

Country Status (2)

Country Link
CN (1) CN106257447B (zh)
WO (1) WO2016201992A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112785400A (zh) * 2021-01-12 2021-05-11 四川天行健穗金科技有限公司 一种用于去财税数据的智能检索方法及系统

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108965906B (zh) * 2018-05-28 2020-12-29 安徽维德工业自动化有限公司 一种基于云存储服务器的视频存储方法
CN109151557B (zh) * 2018-08-10 2021-02-19 Oppo广东移动通信有限公司 视频创建方法及相关装置
CN109271552B (zh) * 2018-08-22 2021-08-20 北京达佳互联信息技术有限公司 通过图片检索视频的方法、装置、电子设备及存储介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102663005A (zh) * 2012-03-19 2012-09-12 杭州海康威视系统技术有限公司 基于云计算的海量视频文件存储系统、分析方法及其系统
CN103365854A (zh) * 2012-03-28 2013-10-23 鸿富锦精密工业（深圳）有限公司 视频文件检索系统及检索方法
CN103473327A (zh) * 2013-09-13 2013-12-25 广东图图搜网络科技有限公司 图像检索方法与系统
US20150127626A1 (en) * 2013-11-07 2015-05-07 Samsung Tachwin Co., Ltd. Video search system and method

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN100556114C (zh) * 2006-11-23 2009-10-28 中兴通讯股份有限公司 一种媒体分片存储的方法
CN101753271B (zh) * 2009-12-29 2013-05-08 中兴通讯股份有限公司 多描述编码的分布式媒体流分发系统及方法
CN102193923A (zh) * 2010-03-05 2011-09-21 杭州华三通信技术有限公司 一种基于计算机程序的专利信息获取方法和装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102663005A (zh) * 2012-03-19 2012-09-12 杭州海康威视系统技术有限公司 基于云计算的海量视频文件存储系统、分析方法及其系统
CN103365854A (zh) * 2012-03-28 2013-10-23 鸿富锦精密工业（深圳）有限公司 视频文件检索系统及检索方法
CN103473327A (zh) * 2013-09-13 2013-12-25 广东图图搜网络科技有限公司 图像检索方法与系统
US20150127626A1 (en) * 2013-11-07 2015-05-07 Samsung Tachwin Co., Ltd. Video search system and method

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112785400A (zh) * 2021-01-12 2021-05-11 四川天行健穗金科技有限公司 一种用于去财税数据的智能检索方法及系统

Also Published As

Publication number Publication date
CN106257447A (zh) 2016-12-28
CN106257447B (zh) 2019-11-26

Similar Documents

Publication Publication Date Title
US11836533B2 (en) Automated reconfiguration of real time data stream processing
US20220070065A1 (en) Enriched flow data for network analytics
Hu et al. Toward an SDN-enabled big data platform for social TV analytics
Chang et al. Bringing the cloud to the edge
Zhang et al. A video cloud platform combing online and offline cloud computing technologies
WO2016201992A1 (zh) 云存储服务器的视频存储及检索方法、视频云存储系统
US10594542B2 (en) System and method for network root cause analysis
US10341277B2 (en) Providing video to subscribers of a messaging system
US10447623B2 (en) Data storage systems and methods using a real-time messaging system
US10698863B2 (en) Method and apparatus for clearing data in cloud storage system
CN103260050A (zh) 基于Google App Engine云平台的视频点播系统
US9197716B2 (en) Pre-fetching resources by predicting user actions
US11055223B2 (en) Efficient cache warm up based on user requests
CN114528104A (zh) 任务处理方法及装置
Zamani et al. Submarine: A subscription‐based data streaming framework for integrating large facilities and advanced cyberinfrastructure
Nirmala Wan optimization tools, techniques and research issues for cloud-based big data analytics
US9648103B2 (en) Non-uniform file access in a distributed file system
WO2024027288A1 (zh) 资源渲染方法、装置、设备、计算机可读存储介质及计算机程序产品
CN109743585B (zh) 一种监控视频收藏以及收藏夹克隆的方法和装置
US10572486B2 (en) Data communication in a distributed data grid
CN103442034A (zh) 一种基于云计算技术的流媒体服务方法和系统
Zhang et al. Processing geo-dispersed big data in an advanced mapreduce framework
Zhang et al. Towards a high speed video cloud based on batch processing integrated with fast processing
CN113810475B (zh) 一种基于大数据架构的Wifi探针设备管控系统
CN111858260A (zh) 信息显示方法、装置、设备及介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 16810733

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 16810733

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 150618)

122 Ep: pct application non-entry in european phase

Ref document number: 16810733

Country of ref document: EP

Kind code of ref document: A1