WO2021036832A1 - 网络摄像机、视频监控系统及方法 - Google Patents

网络摄像机、视频监控系统及方法 Download PDF

Info

Publication number
WO2021036832A1
WO2021036832A1 PCT/CN2020/109328 CN2020109328W WO2021036832A1 WO 2021036832 A1 WO2021036832 A1 WO 2021036832A1 CN 2020109328 W CN2020109328 W CN 2020109328W WO 2021036832 A1 WO2021036832 A1 WO 2021036832A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
face
storage unit
identification number
similarity
Prior art date
Application number
PCT/CN2020/109328
Other languages
English (en)
French (fr)
Inventor
李文伟
徐鹏
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
Priority to EP20857136.4A priority Critical patent/EP4024275A4/en
Priority to US17/753,223 priority patent/US11750776B2/en
Publication of WO2021036832A1 publication Critical patent/WO2021036832A1/zh

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N7/00—Television systems
    • H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
    • H04N7/183—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast for receiving images from a single remote source
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F7/00—Methods or arrangements for processing data by operating upon the order or content of the data handled
    • G06F7/22—Arrangements for sorting or merging computer data on continuous record carriers, e.g. tape, drum, disc
    • G06F7/24—Sorting, i.e. extracting data from one or more carriers, rearranging the data in numerical or other ordered sequence, and rerecording the sorted data on the original carrier or on a different carrier or set of carriers sorting methods in general
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/761—Proximity, similarity or dissimilarity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/98—Detection or correction of errors, e.g. by rescanning the pattern or by human intervention; Evaluation of the quality of the acquired patterns
    • G06V10/993—Evaluation of the quality of the acquired pattern
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/172—Classification, e.g. identification
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
    • H04N23/60—Control of cameras or camera modules
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
    • H04N23/60—Control of cameras or camera modules
    • H04N23/66—Remote control of cameras or camera parts, e.g. by remote control devices
    • H04N23/661—Transmitting camera control signals through networks, e.g. control via the Internet
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
    • H04N23/80—Camera processing pipelines; Components thereof
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N7/00—Television systems
    • H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast

Definitions

  • This application relates to the technical field of video surveillance, and in particular to a network camera, a video surveillance system and a method.
  • the face image is collected by the network camera of the video surveillance system and sent to the back-end server, and the back-end server performs face comparison.
  • the facial images collected by the network cameras cannot be used in some rear-end servers. The problem of successful face comparison or low accuracy of the comparison result on the end server.
  • the first aspect of the present application provides a network camera, including: an image sensor, a processor, a memory, and a network communication interface;
  • the memory includes: a first storage unit and a second storage unit; the image sensor is used to collect monitoring scenes The image; the first storage unit is used to store all the images collected by the image sensor, the second storage unit is used to store part of the images collected by the image sensor, the partial images are the A subset of all images;
  • the processor is configured to execute: match the current image collected by the image sensor with the image stored in the second storage unit to obtain a similarity value representing the matching result, and compare all The similarity value is compared with the first similarity threshold and the second similarity threshold indicated by the outside; when the similarity value is less than the first similarity threshold but greater than the second similarity threshold, pass the The network communication interface sends the current image to the server with a matching failure message; when the similarity value is less than the first similarity threshold and less than the second similarity threshold, the network communication interface is used to The matching failure message send
  • a second aspect of the present application provides a video surveillance system, including: at least one network camera and a server, each network camera and the server establish a communication connection through the network; for any one of the at least one network device, the The network camera includes: an image sensor, a first processor, a first memory, and a first network communication interface; the first memory includes: a first storage unit and a second storage unit; the image sensor is used to collect monitoring scenes The image; the first storage unit is used to store all the images collected by the image sensor, the second storage unit is used to store part of the images collected by the image sensor, the partial images are the A subset of all images; the first processor is configured to execute: match the current image collected by the image sensor with the image stored in the second storage unit to obtain a similarity value representing the matching result , Comparing the similarity value with a first similarity threshold and a second similarity threshold indicated by the outside world, and when the similarity value is less than the first similarity threshold but greater than the second similarity threshold, Send the current image to the server as a
  • a third aspect of the present application provides a method for updating a face database in a camera, including: obtaining a face capture image in response to a face capture instruction; and matching the face according to the face database stored locally by the camera
  • the captured image and any face image in the face database are calculated, and the similarity value of the captured image of the face is calculated; wherein, the face database stores at least two face images, and each face image uniquely corresponds to An identification number; wherein the identification number is used to indicate the time when the corresponding face image is stored in the face database; and the identification number corresponds to a frequency value, and the frequency value is used to indicate the identification
  • a fourth aspect of the present application provides a method for associating a camera with a face database in a server, including: the camera obtains a face capture image in response to a face capture instruction; the camera obtains a face capture image based on the person stored locally by the camera A face database, matching the captured face image with any face image in the face database, and calculating the similarity value of the captured face image; wherein at least two face images are stored in the face database, And each face image uniquely corresponds to an identification number; wherein, the identification number is used to indicate the time when the corresponding face image is stored in the face database in the camera; and the identification number corresponds to a frequency value, The frequency value is used to indicate the number of successful matching of the face image corresponding to the identification number; when the similarity value of the captured image of the face is greater than the first threshold, the camera determines that the matching is successful, and obtains the The face image with the highest similarity value to the captured face image in the face database and the first identification number corresponding to the face image; sending the first identification
  • the images collected by the network camera in the embodiment of the application are processed by its own processor, and the collected images are stored in a targeted manner according to the processed results, which not only reduces the difficulty of image comparison, but also improves the accuracy of the comparison result.
  • the network camera sends the result of image processing to the server, without the need for the server to perform image comparison, which solves the problem that the face image collected by the network camera cannot be successfully compared on some back-end servers or the accuracy of the comparison result is low. .
  • FIG. 1 is a schematic diagram of the structure of a network camera provided by an embodiment of the application
  • FIG. 2 is a schematic diagram of a flow of storing captured images in a second storage unit by the network camera in the embodiment shown in FIG. 1;
  • FIG. 3 is a schematic diagram of an implementation manner in which the network camera uploads captured images to the server in the embodiment shown in FIG. 1;
  • Fig. 4 is a diagram of the attribute setting interface of the second storage unit
  • FIG. 5 is a schematic diagram of an implementation manner of image deletion in the second storage unit
  • FIG. 6 is a schematic diagram of detailed information of the target image in the second storage unit displayed by the display device
  • FIG. 7 is a schematic diagram of the design of the human-computer interaction interface of the display device.
  • FIG. 8 is a schematic diagram of the structure of the video surveillance system provided by the implementation of this application.
  • FIG. 9 is a schematic flowchart of Embodiment 1 of a method for updating a face database in a camera provided by this application;
  • Embodiment 10 is a schematic flowchart of Embodiment 2 of the method for updating a face database in a camera provided by this application;
  • FIG. 11 is a schematic flowchart of an embodiment of a method for associating a face database in a camera and a server provided by this application.
  • Fig. 1 is a schematic structural diagram of a network camera provided by an embodiment of the present application.
  • the network camera provided by the embodiment of the present application may include: an image sensor 11, a processor 12, a memory 13 and a network communication interface 14.
  • the memory 13 includes: a first storage unit 131 and a second storage unit 132.
  • the image sensor 11 is used to collect images of the surveillance scene
  • the first storage unit 131 is used to store all the images collected by the image sensor 11, and the second storage unit 132 is used to store part of the images collected by the image sensor 11, which is a subset of all the above-mentioned images.
  • the processor 12 is configured to match the current image collected by the image sensor 11 with the image stored in the second storage unit 132 to obtain a similarity value representing the matching result, and compare the similarity value with that indicated by the outside world.
  • the first similarity threshold is compared with the second similarity threshold; when the similarity value is less than the first similarity threshold but greater than the second similarity threshold, the current image is sent to the server with a matching failure message through the network communication interface 14 , And when the similarity value is less than the first similarity threshold and less than the second similarity threshold, the current image is sent to the server with a matching failure message through the network communication interface 14, and a uniquely associated image is assigned to the current image.
  • the device target is identified, and the current image is stored in the second storage unit 132.
  • the first similarity threshold is greater than the second similarity threshold, and the device target identification of the current image is generated based on the time when the current image is stored in the second storage unit 132, the identification number of the network camera, and a random value.
  • the first similarity threshold and the second similarity threshold indicated by the outside world refer to the preset first similarity threshold and second similarity threshold.
  • the first similarity threshold and the second similarity threshold can also be set according to user input. Order to make adjustments.
  • the first storage unit 131 in the embodiment of the present application is used to store the image sensor. All images collected by 11 mean that all images collected by the image sensor 11 will be sent to the first storage unit 131 for storage. When the first storage unit 131 is full, the earliest stored image can be overwritten. The first storage unit 131 and the like can be replaced with a new one.
  • the matching failure message and the following matching success message can be uploaded through fields of different protocols, so that the server can distinguish the matching result when the network camera is uploaded.
  • a designated field can be set in the header of the message to distinguish matching results.
  • the designated field in the matching failure message is the first character (for example, 0), and the designated field in the matching success message is the second character (for example, 1).
  • the matching result can be distinguished by identifying the characters in the designated field.
  • the network camera uses the image sensor 11 to collect images of the monitoring scene where it is located.
  • the image sensor 11 is used to capture the image containing the The current image of the face.
  • the image sensor 11 transmits the current image to the first storage unit 131 for storage.
  • the processor 12 uses face detection technology to analyze the facial features in the current image and the second storage unit 132.
  • the facial features of the stored image in the current image are matched with the image stored in the second storage unit 132 to obtain a matching result.
  • the matching result includes: the similarity value between the current image and the above-mentioned stored image .
  • the similarity value between the current image and the stored image may include all the similarity values between the current image and all the stored images in the second storage unit 132, or it may be the maximum similarity value among all the similarity values. Value, the embodiment of the present application does not limit it, and it can be customized according to the actual situation.
  • the processor 12 first matches the similarity value with a first similarity threshold indicated by the outside world, and determines a matching result.
  • the first similarity threshold is used to determine whether the current image matches the images in all the images in the second storage unit 132 successfully.
  • the similarity values between the current image and the images stored in the second storage unit 132 are all less than the first similarity threshold, it indicates that there is no content in the second storage unit 132 (content in this application).
  • the target of interest for example, it can be an image that is consistent with the current image content, for example, it can be a face, a human body, or a vehicle.
  • the network communication interface 14 can be used to send the current image to the server through a matching failure message.
  • the matching failure message is used to indicate that the current image has failed to match in the network camera, that is, there is no previous image captured by the network camera. Images with consistent content so that the server can perform corresponding processing.
  • the processor 12 may also match the similarity value with a second similarity threshold indicated by the outside world to determine whether the current image satisfies the storage to the first 2. Conditions of the storage unit 132.
  • the second similarity threshold is used to determine whether to store the current image in the second storage unit 132.
  • the first similarity threshold is greater than the second similarity threshold, that is, when the matching result is that the matching is successful, the operation of storing the current image in the second storage unit 132 will not be performed. Therefore, there is no need to perform the judgment whether to The current image is stored in the second storage unit 132, which reduces the amount of calculation.
  • a unique associated device target identifier may be assigned to the current image first, and the device target identifier of the current image may be stored in the second storage unit 132 according to the current image.
  • the time, the identification number of the network camera, and the random value are generated, so as not only to ensure that the generated device target identification is unique, but also to identify when the current image was taken by which network camera.
  • the device target identifier of the image may be a unique identifier used to indicate the image inside the device, which may use 64-bit characters, and the specific composition method may be the device serial number + storage time + random number. This application does not limit the specific composition of the device target identifier, which can be determined according to actual conditions.
  • FIG. 2 is a schematic diagram of a flow of storing captured images in the second storage unit by the network camera in the embodiment shown in FIG. 1.
  • the image storage process includes the following steps: S21, the image sensor collects the current image; S22, the processor determines whether the similarity value between the current image and the image in the second storage unit is less than the first value indicated by the outside world.
  • Similarity threshold if not, execute S23, send the current image to the server through the network communication interface with a matching success message; if yes, execute S24, send the current image to the server through the network communication interface with a matching failure message, S25, judge Whether the similarity value is less than the second similarity threshold indicated by the outside world, if yes, execute S26, assign a unique associated device target identifier to the current image, and store the current image in the second storage unit, if not, execute S27, according to External instructions for processing.
  • the similarity value between the current image and the image in the second storage unit in S22 may specifically be: the maximum similarity value between the current image and each image in the second storage unit.
  • the embodiment of the present application does not limit the specific implementation principle of the foregoing S27, which may be determined based on actual conditions.
  • the network camera uses the image sensor to collect the image of the monitoring scene.
  • the collected image is stored in the first storage unit of the memory; on the other hand, the processor is used to collect the current image collected by the image sensor.
  • the image is matched with the image stored in the second storage unit of the memory to obtain a similarity value representing the matching result.
  • the current image is sent to the server with a matching failure message through the network communication interface, and when the similarity value is less than the first similarity threshold and less than the second similarity threshold, the current image is sent to the server with a matching failure message through the network communication interface.
  • a unique associated device target identifier is also assigned to the current image, and the current image is stored in the second storage unit.
  • the processor 12 is specifically configured to obtain the image quality of the current image when it is determined that the similarity value is less than the first similarity threshold and less than the second similarity threshold.
  • the current image is sent to the server with a matching failure message through the network communication interface 14, and a unique associated device target identifier is assigned to the current image, and the current image is stored in the first Two storage unit 132.
  • the processor 12 of the network camera can make a quality judgment before storing the current image in the second storage unit 132, and only store it in the current image when the image quality of the current image is greater than the preset image quality.
  • the second storage unit 132 does not store it otherwise.
  • the specific value of the preset image quality is not limited, and it can be customized according to the actual situation.
  • the maximum value of the image collected by the image sensor 11 can be determined according to the storage space of the second storage unit 132. To determine the best quality, etc., I won’t repeat them here.
  • the quality of the image stored in the second storage unit can be improved by judging the image quality of the current image, and using high-quality images to judge the similarity, the accuracy of the result is also higher, which is more accurate for the follow-up.
  • the processor 12 is further configured to compare the similarity value with a third similarity threshold indicated by the outside world, and when the similarity value is greater than the third similarity threshold, obtain the current The image quality of the image. If the image quality is greater than the image quality of the matching image corresponding to the current image, the current image is used to replace the matching image in the second storage unit 132, and the device target identifier of the matching image is used as the device of the current image The target identification, wherein the third similarity threshold is greater than the first similarity threshold.
  • the third similarity threshold indicated by the outside world refers to a preset third similarity threshold.
  • the third similarity threshold may also be adjusted according to instructions input by the user.
  • the processor 12 may store all the images meeting the second similarity threshold and the preset image quality in the second storage unit 132 when storing the image for the first time.
  • the image quality of the current image captured may be greater than the original matching image stored in the second storage unit 132.
  • the current image and the second storage unit 132 The similarity value of the matching image in the unit 132 is greater than the third similarity threshold.
  • the current image can be used to replace the matching image in the second storage unit 132.
  • the device target identifier of the matching image is used as the device target identifier of the current image at this time.
  • the third similarity threshold should be greater than the first similarity threshold.
  • the third similarity threshold may be 88%, Values such as 90%, the embodiment of the application does not limit the specific values of the first similarity threshold and the third similarity threshold, as long as the condition that the third similarity threshold is greater than the first similarity threshold is satisfied.
  • the image in the second storage unit has an update mechanism, which can further reduce the processing pressure of the network camera and improve the accuracy of the image comparison result.
  • the aforementioned processor 12 is further configured to determine the current storage capacity of the second storage unit 132 and the first ratio of the image in the second storage unit 132, according to the current storage capacity, The first ratio and the image storage upper limit value of the second storage unit 132 indicated by the outside world are adjusted, and the second similarity threshold is adjusted.
  • the first ratio is used to indicate the ratio of the number of target objects corresponding to the current storage capacity to the current storage capacity.
  • both the first similarity threshold and the second similarity threshold may be self-defined values, where the second similarity threshold may be associated with the storage capacity of the second storage unit 132.
  • the processor 12 may dynamically adjust the second degree of similarity based on the storage capacity of the second storage unit 132, the first ratio of the image in the second storage unit 132, and the image storage upper limit value of the second storage unit 132 indicated by the outside world. Threshold, so that the number of images stored in the second storage unit 132 does not exceed the storage upper limit.
  • the first ratio of the image can also be characterized by the storage repetition rate of the image, the accuracy of the image, etc., a high first ratio indicates that the storage accuracy rate is high, but the storage repetition rate is low.
  • the first ratio represents the ratio of the total number of different objects to the total number of images in each image stored in the second storage unit 132. For example, there are 100 images of people stored in the second storage unit 132, but the 100 images actually only correspond to 90 different people, then the first ratio is 90%, that is, the images stored in the second storage unit 132 The image repetition rate is 10%.
  • the image storage upper limit value of the second storage unit 132 indicated by the outside world may specifically be the number of different objects stored in the second storage unit 132.
  • the greater the storage capacity of the second storage unit 132, the higher the second similarity threshold; the higher the first ratio of the images in the second storage unit 132, the higher the second similarity threshold; the second similarity threshold indicated by the outside world The higher the image storage upper limit value of the storage unit 132, the higher the second similarity threshold.
  • the processor 12 is further configured to use the current image to replace the image with the least number of successful matches in the second storage unit when the current storage capacity is equal to the image storage upper limit value.
  • the number of successful matching of each image in the second storage unit 132 can be analyzed to determine the one with the lowest number of successful matching.
  • the image with the lowest number of successful matches may be the image that was mistakenly stored in the second storage unit 132, or the target corresponding to the image does not often appear in the area monitored by the network camera. Therefore, the image in the second storage unit 132
  • the current image may be used to replace the image with the least number of successful matching in the second storage unit 132.
  • the processor 12 is further configured to determine that the current time is the successful time of the match when the similarity value is greater than or equal to the first similarity threshold, and judge Whether the time difference between the successful matching time this time and the last successful matching time is greater than or equal to the preset time difference, if so, the current image is sent to the server through the network communication interface 14 as a matching success message, and the last successful matching time Update the time of successful matching, and add 1 to the number of successful matching of the matching image corresponding to the current image. If not, discard the current image.
  • the similarity value is greater than or equal to the first similarity threshold, it indicates that there is an image with content consistent with the current image content in the second storage unit 132.
  • the target matching image is the similarity value of the current image in the second storage unit 132 The largest image.
  • the processor 12 can upload the current image to the server. Specifically, it can send a matching success message to the server through the network communication interface 14.
  • the matching success message is used to indicate that the current image is successfully matched in the network camera, that is, There is an image with the same content in the image taken by the network camera before, which can indicate that the network camera has uploaded the current image before; on the other hand, in the network camera, the processor 12 can also match the current image in the second storage unit 132 The number of successful matching times is increased by 1, so that the number of successful matching times of matching images in the second storage unit 132 is counted.
  • the matching image corresponding to the current image matching refers to the matching image that is successfully matched to the current image. The more matching images of the matching image corresponding to the current image matching, the more it indicates that the content of the current image is frequently collected content.
  • the time difference between the successful matching time this time and the successful matching time last time is less than the preset time difference, it indicates that the image sensor 11 repeats sampling within the preset time period, and the current image may not be processed at this time. , Or just discard it.
  • the preset time difference may be 5s, 8s, 10s, etc.
  • the accuracy of the preset time difference may be For 0.1s and so on.
  • FIG. 3 is a schematic diagram of an implementation manner in which the network camera uploads captured images to the server in the embodiment shown in FIG. 1.
  • the implementation may include the following steps: S31, the image sensor collects the current image; S32, the processor determines whether the similarity value between the current image and the image in the second storage unit is less than the first value indicated by the outside world.
  • Similarity threshold if yes, execute S33, send the current image to the server with a matching failure message through the network communication interface; if not, execute S34, determine that the current time is the matching success time, S35, determine the matching success time and Whether the time difference between the last successful matching time is greater than or equal to the preset time difference, if not, execute S36, discard the current image; if yes, execute S37, send the current image to the server through the network communication interface with a matching success message, S38, The last successful matching time is updated to the current successful matching time, and the number of successful matching of the current image corresponding to the matching image is increased by 1.
  • the similarity value between the current image and the image in the second storage unit in S32 may specifically be the maximum similarity value between the current image and each image in the second storage unit.
  • the foregoing processor 12 is further configured to periodically obtain the number of successful matching of each image in the second storage unit 132 within a preset time period, and keep the number of successful matching greater than or equal to the external Images with the indicated threshold of the number of successful matching times are deleted, and images with the number of successful matching times less than the threshold of successful matching times indicated by the outside world are deleted.
  • the threshold for the number of successful matches indicated by the outside world may be a preset value or a value set by the user.
  • the threshold for the number of successful matches indicated by the outside world is positively correlated with the preset time period. For example, when the preset time period is 1 day, the threshold for the number of successful matches can be 1, 2, or 3 times; when the preset time period is 1 week, match The threshold of the number of successes can be 5, 10, 20, and so on.
  • the network camera also has the function of periodically deleting the images in the second storage unit 132 to ensure that the images stored in the second storage unit 132 are frequently used, thereby improving the use of the images stored in the second storage unit 132. frequency.
  • functions such as image addition, deletion, and attribute setting for the second storage unit 132 can be completely compatible with the management of the storage unit under normal circumstances, and it can receive attribute values specified by the user through a human-computer interaction interface.
  • FIG. 4 is a diagram of the attribute setting interface of the second storage unit.
  • the information of the second storage unit may specifically be as follows:
  • second storage unit 1 first similarity threshold (that is, matching success threshold): 85%, remarks: none;
  • Open storage not open (where the box is unchecked means it is not turned on, and the box is checked means it is turned on), the second similarity threshold (that is, the storage similarity threshold): 70%, the image quality threshold: 60%;
  • FIG. 5 is a schematic diagram of an implementation manner of image deletion in the second storage unit.
  • the implementation manner may include the following steps: S51. Turn on automatic deletion; S52.
  • the processor periodically obtains the number of successful matching of each image in the second storage unit within a preset time period; S53. For each image in the second storage unit, determine whether the number of successful matching times of the image is greater than or equal to the threshold of successful matching times indicated by the outside world; if yes, execute S54 to retain the image; if not, execute S55 to delete the image.
  • the aforementioned processor 12 is further configured to, in response to an external image ordering instruction, order the images in the second storage unit 132 according to the image ordering instruction.
  • the sorting result is obtained.
  • the image sorting indication is used to indicate the sorting method of the images in the second storage unit 132.
  • the sorting method is based on parameters: number of successful matching, successful matching time, storage time, etc., which can be used according to the size of the parameter Sort from largest to smallest or from smallest to largest.
  • the processor 12 may sort the images in the second storage unit 132 based on an external image sorting instruction to obtain the sorting result, so that the processor 12 may transmit the sorting result to the network camera connected to the network camera.
  • a display device so that the display device presents the above sorting result.
  • the display device may have a human-computer interaction interface that can be manipulated by the user, that is, the user can input or generate an image ordering instruction through the human-computer interaction interface to indicate that the second storage unit 132 In what sort of order the images can be displayed.
  • the network camera can provide at least any one of the number of successful matching, the time of successful matching, and the storage time for each image. put in order.
  • the above-mentioned processor 12 is further configured to respond to a display instruction for the target image issued by the outside, and based on the display instruction, obtain the similarity between the target image and the target image from the first storage unit 131.
  • the multiple similar images with a degree greater than the first similarity threshold are used to display the target image and the multiple similar images.
  • the processor 12 may store all the images captured by the image sensor 11. Since the images stored in the second storage unit 132 are a subset of the images stored in the first storage unit 131, specifically, the first The first storage unit 131 stores all the captured images, and the second storage unit 132 stores all the above-mentioned images that meet the second similarity threshold. Therefore, the images in the first storage unit 131 and the second storage The images in the unit 132 have a certain corresponding relationship. For example, the image in the first storage unit 131 can carry the identifier of the matching image and has the time when the matching is successful. The matching image is the image in the second storage unit 132.
  • the processor 12 may respond to the display instruction and obtain a preset number of similar images from the first storage space based on the identification of the target image in the display instruction and the above-mentioned corresponding relationship, so as to achieve the target Display of images and multiple similar images.
  • the network camera may be connected to a display device, and thus, the display device may display images in the first storage unit and/or the second storage unit in the network camera.
  • the display device has a human-computer interaction interface, so that when the human-computer interaction interface of the display device presents an image in the second storage unit 132, the user can click on an image in the second storage unit 132 through the human-computer interaction interface , The detailed information of the image can be displayed on the human-computer interaction interface.
  • FIG. 6 is a schematic diagram of detailed information of the target image in the second storage unit displayed by the display device.
  • the human-computer interaction interface of the display device may include preview, recycling (ie, delete), picture, application, configuration and other functional options.
  • the display interface of the second storage unit has filtering options. Conditions, etc.
  • the human-computer interaction interface can also present detailed information of the image, that is, the image details.
  • image attributes may include information such as the name, gender, province, city, second similarity threshold, storage time, and total number of successful matching of the person corresponding to the image.
  • the shooting record may include: matching time, acquaintance, image quality, and so on.
  • FIG. 7 is a schematic diagram of the design of the human-computer interaction interface of the display device.
  • the human-computer interaction interface can be divided into multiple areas, such as preview area, matching result, snapshot display, analysis area, second storage unit, deleted image display, and so on. The specific distribution of each area is shown in Figure 7, and will not be repeated here.
  • the network camera provided by the embodiments of the present application can process the images taken in the surveillance scene, automatically maintain the image in the second storage unit, reduce the repeated processing process, increase the degree of automation of the network camera, and improve the image contrast. The accuracy of the result.
  • the network camera can assign a unique associated device target identifier to each image generated, and the image transmitted to the server also carries the device target identifier, so that accurate images can be obtained without the need for the server to perform image matching. compare results.
  • FIG. 8 is a schematic structural diagram of a video surveillance system provided by the implementation of this application.
  • the video surveillance system may include: at least one network camera (for example, network camera 81 to network camera 8n, where n is a positive integer) and a server 80, and each network camera and server 80 establish communication through the network connection.
  • Each network camera in the above at least one network camera may have the same configuration, and the implementation principle of each network camera is similar.
  • one network camera 81 of the above at least one network camera is used for explanation.
  • the network camera 81 includes: an image sensor 811, a first processor 812, a first memory 813, and a first network communication interface 814.
  • the first memory 813 includes a first storage unit 8131 and a second storage unit 8131. Unit 8132.
  • the image sensor 811 is used to collect images of the monitoring scene
  • the first storage unit 8131 is used to store all the images collected by the image sensor 811, and the second storage unit 8132 is used to store part of the images collected by the image sensor 811, and this part of the image is a child of all the images. set;
  • the first processor 812 is configured to execute:
  • the current image collected by the image sensor 811 is matched with the image stored in the second storage unit 8132 to obtain a similarity value representing the matching result, and the similarity value is compared with the first similarity indicated by the outside world
  • the threshold and the second similarity threshold are compared, and when the similarity value is less than the first similarity threshold but greater than the second similarity threshold, the current image is sent to the server through the first network communication interface 814 with a matching failure message 80, and when the similarity value is less than the first similarity threshold and less than the second similarity threshold, the current image is sent to the server 80 as a matching failure message through the first network communication interface 814, and a uniquely associated image is assigned to the current image.
  • the device target identifier, and the current image is stored in the second storage unit 8132.
  • the first similarity threshold is greater than the second similarity threshold
  • the device target identification of the current image is based on the time when the current image is stored in the second storage unit 8132, and the identification of the network camera 81 Number and random value.
  • the server 80 includes: a second network communication interface 821, a second processor 822, and a second memory 823.
  • the second processor 822 is configured to receive the matching failure message sent by the network camera through the second network communication interface 821, and determine whether there is an image consistent with the device target identifier of the current image in the second memory 823, and if not, it is The current image is assigned a uniquely associated platform target identifier, and the corresponding relationship between the device target identifier and the matching failure message is established, and the platform target identifier and the corresponding relationship are stored; if so, the current image is acquired in the second memory 823
  • the platform target identifier of establishes the mapping relationship between the device target identifier and the platform target identifier, and stores the current image and the mapping relationship in the second memory 823.
  • the server 80 may receive images sent by multiple network cameras through the second network communication interface 821, and this embodiment is explained by receiving images sent by the above-mentioned network camera 81.
  • the server 80 may communicate through the second network accordingly.
  • the interface 821 receives the current image.
  • the second processor 822 of the server 80 may determine that the network camera may not have sent a matching image of the current image before, but In order to avoid the missing record of the matching image of the current image, the second processor 822 may perform similarity matching on the current image, and determine whether the matching image of the current image is stored in the server 80, that is, determine whether the second memory 823 There is an image consistent with the device target identifier of the current image, and corresponding processing is executed according to the judgment result.
  • the corresponding relationship between the device target identifier and the matching failure message is established to record that the content of the current image is not recorded on the server 80 In order to make a mark so that relevant personnel can deal with it in the follow-up.
  • the platform target identifier of the matching image corresponding to the current image in the second memory 823 is acquired, and the device target identifier and the platform are established.
  • the mapping relationship of the target identifier is stored and stored in the second memory 823.
  • the second processor 822 can acquire the matching image.
  • the platform target identifier in the second storage 823 establishes and stores the mapping relationship between the device target identifier of the current image and the platform target identifier, so that when the network camera uploads the matching image of the current image again, it can directly obtain one of the two.
  • the mapping relationship between the two is directly mapped to the previous analysis result, which reduces the computing power occupied by the server 80 and reduces the processing burden of the server 80.
  • the platform target identifier in this embodiment may be the detailed information of the content of the image in the server, for example, the actual identity information of the person corresponding to the face in the face image, for example, it may be an identity identifier such as ID or name. information.
  • This application does not limit the content of the platform target identifier, which can be set according to actual needs.
  • the first processor 812 is further configured to use the first network communication interface 814 to match successfully when the similarity value is greater than or equal to the first similarity threshold.
  • the message sends the current image to the server 80;
  • the second processor 822 is further configured to receive the matching success message sent by the network camera through the second network communication interface 821, obtain the platform target identifier of the matching image corresponding to the current image in the second memory 823, and the platform target The identified analysis result is used as the analysis result of the current image.
  • the second processor 822 of the server 80 can determine that the network camera may have sent a matching image of the current image before, and then directly Obtain the platform target identifier of the matching image corresponding to the current image in the second memory 823, and use the analysis result of the platform target identifier as the analysis result of the current image. There is no need to analyze the current image again, which reduces the computing power occupied by the server 80. The processing tasks of the server 80 are reduced, and the processing burden of the server 80 is reduced.
  • the second processor 822 is further configured to send the platform target identifier of the current image to the network camera through the second network communication interface 821;
  • the first processor 812 is further configured to receive the platform target identifier of the current image through the first network communication interface 814, and the image obtained by the image sensor 811 is similar to the image stored in the second storage unit 8132.
  • the degree value is greater than or equal to the first similarity threshold, send the platform target identifier as the device target identifier of the image to the server 80;
  • the second processor 822 is further configured to directly obtain the analysis result of the image according to the platform target identifier of the received image.
  • the first processor 812 uploads the current image with a matching success message, it indicates that the matching image of the current image is stored in the second storage unit 8132 of the network camera.
  • the second processor 822 can acquire The platform target identifier with the current image is sent to the network camera.
  • the network camera After the network camera receives the platform target identifier, on the one hand, it can directly use the platform target identifier as the device target identifier of the matching image corresponding to the current image, so that the network camera can use the platform target identifier as the device target identifier when uploading the matched image again
  • the second processor 822 of the server 80 can directly obtain the analysis result of the image according to the platform target identifier, without obtaining the platform target identifier that has a mapping relationship with the device target identifier, which reduces the workload of searching for the mapping relationship.
  • the network camera can maintain the mapping relationship between the platform target ID and the device target ID.
  • the platform target ID can be determined first, and then the platform target ID can be uploaded, which can also reduce the mapping of the server 80.
  • the workload of the relationship search reduces the processing burden of the server 80.
  • the network camera can perform self-maintenance on the image in the second storage unit of the current point, and form a unique second storage unit for the current point through long-term accumulation.
  • the end server forms a mapping relationship between the device target ID of the image in the device and the platform target ID of the image in the back-end large storage. Once the subsequent received images are matched successfully, the server does not need to match again, but matches the network camera successfully.
  • the device target identification of the image is directly mapped, which not only reduces the processing pressure, but also avoids the problems of incompatibility between the front-end network camera and the server and the difficulty of image maintenance in the second storage unit.
  • FIG. 9 is a schematic flowchart of Embodiment 1 of a method for updating a face database in a camera provided by this application.
  • the method may include the following steps:
  • the camera can perform a face capture image acquisition operation under an external trigger.
  • the camera may obtain a face capture instruction indicated by the outside world, and according to the face capture instruction, the image sensor of the camera is used to collect a face capture image.
  • the processor of the camera captures at least one face captured The image is processed.
  • the face database stores at least two face images, and each face image uniquely corresponds to an identification number; wherein, the identification number is used to indicate the time when the corresponding face image is stored in the face database; and The identification number corresponds to a frequency value, and the frequency value is used to indicate the number of successful matching of the face image corresponding to the identification number.
  • the identification number may include the serial identification number of the camera, a random value, and the time when the corresponding face image is stored in the face database.
  • the camera locally maintains a face database, and the face database stores part of the images captured by the camera.
  • the face database stores part of the images captured by the camera.
  • at least two face images in the face database are obtained by collecting faces of different users.
  • each face image uniquely corresponds to an identification number and a frequency value.
  • the camera when the camera acquires a new captured face image, it can match the captured face image with the face image in the local face database to determine the similarity value of the captured face image to determine whether The face database needs to be updated.
  • the similarity value of the captured face image when it is determined that the similarity value of the captured face image is greater than the first threshold, it indicates that there is a face image in the face database that is basically consistent with the content of the captured face image. Therefore, in order to avoid the same content If the image is stored repeatedly, it is necessary to determine the face image with the highest similarity value to the captured image of the face in the face database, and accordingly, determine the first identification number corresponding to the face image with the highest similarity value and The frequency value corresponding to the first identification number.
  • the method further includes: sending the captured image of the human face and the first identification number to the server.
  • the server can determine the previous processing result of the matching image corresponding to the captured face image according to the identification number, which simplifies the processing operation of the server.
  • the preset threshold corresponding to the frequency value may be used to indicate the determination condition of whether the face image is allowed to be stored in the face database.
  • the frequency value corresponding to the first identification number is less than the preset threshold, it indicates that the face image may be an image that has been mistakenly stored in the face database or that the target corresponding to the image does not often appear in the area where the camera is located.
  • the face image corresponding to the first identification number in the face database in the camera is deleted, and the face database in the camera is updated, so as to ensure the high accuracy rate of the image stored in the face database.
  • step S94 is triggered only after the camera has been running for a period of time, or is aimed at scenarios where the remaining storage space of the face database is insufficient, that is, less than the set capacity threshold.
  • S94 when the frequency value corresponding to the first identification number is less than a preset threshold, delete the face image corresponding to the first identification number in the face database in the camera, and update the face database in the camera, including: When the storage time of the face image corresponding to the first identification number is greater than the preset time and/or the remaining storage space of the face database in the camera is less than the preset capacity threshold, when the frequency corresponding to the first identification number is When the value is less than the preset threshold, the face image corresponding to the first identification number in the face database in the camera is deleted, and the face database in the camera is updated.
  • the method may also perform the following operations first, and then perform S94. That is, after S93, add 1 to the frequency value corresponding to the first identification number, and record the frequency value after the calculation, so as to update the frequency value corresponding to the first identification number.
  • the face image corresponding to the first identification number in the face database of the camera is deleted.
  • Comparing the calculated frequency value with the preset threshold is beneficial to improve the accuracy of determining whether the image stored in the human database is accurate.
  • the method provided by the embodiment of this application obtains a face capture image in response to a face capture instruction, matches the captured face image with any face image in the face database according to the face database stored locally by the camera, and calculates the person The similarity value of the captured face image; when the similarity value of the captured face image is greater than the first threshold, it is determined that the matching is successful, and the face image and the face image with the highest similarity value to the captured face image in the face database are obtained Corresponding to the first identification number and the frequency value corresponding to the first identification number; when the frequency value corresponding to the first identification number is less than the preset threshold, delete the face image corresponding to the first identification number in the face database of the camera, and update the The face database in the camera ensures the accuracy of the images stored in the face database and improves the accuracy of the image comparison results.
  • FIG. 10 is a schematic flowchart of Embodiment 2 of a method for updating a face database in a camera provided by this application. As shown in FIG. 10, in this embodiment, the method may further include the following steps:
  • S101 can be executed after S92, that is, when the similarity value of the captured face image is less than the first threshold, the similarity value of the captured face image is compared with the size of the second threshold. Perform corresponding operations based on the comparison results.
  • S102 Store the captured face image in the face database in the camera, record the time when the captured face image is stored in the face database in the camera, and assign a unique identification number to the captured face image.
  • the second threshold is less than or equal to the first threshold.
  • the similarity value of the captured face image when the similarity value of the captured face image is less than the second threshold, it indicates that there is no image in the face database whose content is consistent with the content of the captured face image.
  • the captured face image can be stored in The face database in the camera.
  • S103 Do not store the captured face image in the face database in the camera, but send the captured face image to the server.
  • the similarity value of the captured face image when the similarity value of the captured face image is greater than or equal to the second threshold, it indicates that an image with content consistent with the content of the captured face image already exists in the face database, in order to avoid repeated storage of the face image , The captured face image will not be stored in the face database in the camera, but the captured face image will be sent to the server so that the server can perform corresponding processing on the captured face image.
  • the similarity value of the captured face image when the similarity value of the captured face image is less than the first threshold, the similarity value of the captured face image is compared with the second threshold, and when the similarity value of the captured face image is less than the second threshold , Store the captured face image to the face database in the camera, record the time when the captured face image is stored in the face database in the camera, and assign a unique identification number to the captured face image.
  • the method may further include the following steps:
  • the method may further include: in response to the received display instruction on the face database, generating a data packet, the data packet is used to display the face database corresponding to the display instruction Face image.
  • the camera can also obtain a display instruction on the face database issued by the outside world, and based on the display instruction, sort and count the images in the face database to generate a data packet so that the camera can
  • the data packet is transmitted to the display device connected to the camera, so that the display device displays the face image corresponding to the display instruction in the face database.
  • the camera can automatically maintain and update the local face database, which reduces the process of repeated image processing, improves the degree of automation of the camera, and improves the accuracy of image comparison results
  • the camera can assign a unique identification number to each face image, and the image transmitted to the server also carries the identification number.
  • the camera in this embodiment is also the network camera in the above embodiment, and the face database in this application is also the second storage unit of the memory in the above embodiment.
  • the face database in this application is also the second storage unit of the memory in the above embodiment.
  • FIG. 11 is a schematic flowchart of an embodiment of a method for associating a face database in a camera and a server provided by this application. As shown in FIG. 11, in this embodiment, the method may include the following steps:
  • S111 The camera acquires a face capture image in response to a face capture instruction.
  • S112 The camera matches the captured face image with any face image in the face database according to the face database stored locally by the camera, and calculates the similarity value of the captured face image.
  • the face database stores at least two face images, and each face image uniquely corresponds to an identification number; wherein, the identification number is used to indicate the time when the corresponding face image is stored in the face database in the camera
  • the identification number corresponds to a frequency value, which is used to indicate the number of successful matching of the face image corresponding to the identification number.
  • S114 The camera sends the first identification number and the captured image of the face to the server;
  • the server compares the captured face image with the face database in the server in response to the first identification number being received for the first time.
  • each face image in the face database in the server uniquely corresponds to a second identification number
  • the server creates an association relationship between the first identification number and the second identification number, and the association relationship is used to associate the face database in the camera with the face database in the server.
  • the method for associating the face database in the camera and the server further includes: when the face image stored in the face database is longer than the first preset time period, and the face image When the number of matching successes corresponding to the identification number is less than the preset threshold, the face image is deleted from the face database. And/or when the remaining storage space of the face database is less than the preset capacity threshold, the number of matching successes corresponding to the identification number in the face database is less than the preset threshold for face images.
  • the camera after the camera acquires a captured face image, it matches the captured face image with any face image in the face database according to the face database stored locally by the camera, and calculates the similarity of the captured face image When the similarity value of the captured face image is greater than the first threshold, it is determined that the matching is successful and the face image with the highest similarity value of the captured face image in the face database and the first identifier corresponding to the face image are obtained And send the first identification number and the captured face image to the server. Correspondingly, according to the received first identification number, the server responds that the first identification number is received for the first time.
  • the face database in the server is compared, and after the face captured image is successfully compared with the face image in the server, the face image and the corresponding second identification number in the server with the highest similarity value to the face captured image are obtained Create an association relationship between the first identification number and the second identification number, and the association relationship is used to associate the face database in the camera with the face database in the server.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Multimedia (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Signal Processing (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Quality & Reliability (AREA)
  • Computing Systems (AREA)
  • Databases & Information Systems (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Human Computer Interaction (AREA)
  • Data Mining & Analysis (AREA)
  • Computer Hardware Design (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Image Analysis (AREA)
  • Studio Devices (AREA)

Abstract

本申请提供一种网络摄像机、视频监控系统及方法，其中，该网络摄像机包括：图像传感器、处理器、存储器和网络通信接口；该处理器用于将图像传感器采集的当前图像与存储器的第二存储单元中已存储的图像进行匹配，得到表示匹配结果的相似度值，将满足相似度条件的图像存储至网络设备摄像的另一个存储单元中，降低了图像对比的难度，提高了对比结果的准确度。

Description

网络摄像机、视频监控系统及方法
本申请要求于2019年08月29日提交中国专利局、申请号为201910808803.2发明名称为“网络摄像机、视频监控系统及方法”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
本申请涉及视频监控技术领域，尤其涉及一种网络摄像机、视频监控系统及方法。
背景技术
通过视频监控系统的网络摄像机采集人脸图像，发送至后端服务器，由后端服务器执行人脸比对。然而，由于市面上各类型的网络摄像机在硬件和软件上存在较大差异，且很多后端服务器在硬件和软加上也存在较大差异，导致网络摄像机采集的人脸图像无法在某些后端服务器上成功执行人脸比对或比对结果准确率低的问题。
发明内容
本申请第一方面提供一种网络摄像机，包括：图像传感器、处理器、存储器和网络通信接口；所述存储器包括：第一存储单元和第二存储单元；所述图像传感器，用于采集监控场景的图像；所述第一存储单元，用于存储所述图像传感器采集到的所有图像，所述第二存储单元，用于存储所述图像传感器采集到的部分图像，所述部分图像为所述所有图像的子集；所述处理器，用于执行：将所述图像传感器采集的当前图像与所述第二存储单元中已存储的图像进行匹配，得到表示匹配结果的相似度值，将所述相似度值与外界指示的第一相似度阈值、第二相似度阈值进行比较；在所述相似度值小于所述第一相似度阈值但大于所述第二相似度阈值时，通过所述网络通信接口以匹配失败消息将所述当前图像发送给所述服务器；在所述相似度值小于所述第一相似度阈值且小于所述第二相似度阈值时，通过所述网络通信接口以匹配失败消息将所述当前图像发送给服务器，且为所述当前图像分配一个唯一关 联的设备目标标识，并将所述当前图像存储至所述第二存储单元；其中，所述第一似度阈值大于所述第二相似度阈值，所述当前图像的设备目标标识是基于所述当前图像被存储至所述第二存储单元的时间、所述网络摄像机的标志号以及随机值生成的。
本申请第二方面提供一种视频监控系统，包括：至少一个网络摄像机和服务器，每个网络摄像机和所述服务器通过网络建立通信连接；对于所述至少一个网络设备中的任意一个网络摄像机，所述网络摄像机包括：图像传感器、第一处理器、第一存储器和第一网络通信接口，所述第一存储器包括：第一存储单元和第二存储单元；所述图像传感器，用于采集监控场景的图像；所述第一存储单元，用于存储所述图像传感器采集到的所有图像，所述第二存储单元，用于存储所述图像传感器采集到的部分图像，所述部分图像为所述所有图像的子集；所述第一处理器，用于执行：将所述图像传感器采集到的当前图像与所述第二存储单元中已存储的图像进行匹配，得到表示匹配结果的相似度值，将所述相似度值与外界指示的第一相似度阈值、第二相似度阈值进行比较，在所述相似度值小于所述第一相似度阈值但大于所述第二相似度阈值时，通过所述第一网络通信接口以匹配失败消息将所述当前图像发送给所述服务器，以及在所述相似度值小于所述第一相似度阈值且小于所述第二相似度阈值时，通过所述网络通信接口以匹配失败消息将所述当前图像发送给服务器，且为所述当前图像分配一个唯一关联的设备目标标识，并将所述当前图像存储至所述第二存储单元，其中，所述第一相似度阈值大于所述第二相似度阈值，所述设备目标标识是基于所述当前图像被存储至所述第二存储单元的时间、所述网络摄像机的标志号以及随机值生成的；其中，所述服务器包括：第二网络通信接口、第二处理器和第二存储器；所述第二处理器，用于通过所述第二网络通信接口接收所述网络摄像机发送的匹配失败消息，判断所述第二存储器是否存在与所述当前图像的设备目标标识一致的图像，若否，为所述当前图像分配一个唯一关联的平台目标标识，并建立所述设备目标标识与匹配失败的对应关系，以及存储所述平台目标标识和所述对应关系；若是，获取所述当前图像对应匹配图像在所述第二存储器中的平台目标标识，建立所述设备目标标识与所述平台目标标识的映射关系，将所 述映射关系存储至所述第二存储器。
本申请第三方面提供一种更新摄像机中人脸数据库的方法，包括：响应于一个人脸抓拍指令，获取一个人脸抓拍图像；根据所述摄像机本地存储的人脸数据库，匹配所述人脸抓拍图像和所述人脸数据库中任一人脸图像，计算所述人脸抓拍图像的相似度值；其中，所述人脸数据库中存储着至少两个人脸图像，且每个人脸图像唯一对应着一个标识号；其中，所述标识号用于指示其对应的人脸图像存储至所述人脸数据库的时间；且所述标识号对应着一个频次数值，所述频次数值用于指示所述标识号对应的人脸图像发生匹配成功的次数；在所述人脸抓拍图像的相似度值大于第一阈值时，确定匹配成功，且获取所述人脸数据库中与所述人脸抓拍图像相似度值最高的人脸图像、所述人脸图像对应的第一标识号以及所述第一标识号对应的频次数值；在所述第一标识号对应的频次数值小于预设阈值时，删除所述摄像机中人脸数据库中所述第一标识号对应的人脸图像，更新所述摄像机中人脸数据库。
本申请第四方面提供一种摄像机和服务器中人脸数据库关联的方法，包括：所述摄像机响应于一个人脸抓拍指令，获取一个人脸抓拍图像；所述摄像机根据所述摄像机本地存储的人脸数据库，匹配所述人脸抓拍图像和所述人脸数据库中任一人脸图像，计算所述人脸抓拍图像的相似度值；其中，所述人脸数据库中存储着至少两个人脸图像，且每个人脸图像唯一对应着一个标识号；其中，所述标识号用于指示其对应的人脸图像存储至所述摄像机中人脸数据库的时间；且所述标识号对应着一个频次数值，所述频次数值用于指示所述标识号对应的人脸图像发生匹配成功的次数；所述摄像机在所述人脸抓拍图像的相似度值大于第一阈值时，确定匹配成功，且获取所述人脸数据库中与所述人脸抓拍图像相似度值最高的人脸图像和所述人脸图像对应的第一标识号；将所述第一标识号和所述人脸抓拍图像发送至所述服务器；所述服务器根据接收到的所述第一标识号，响应于所述第一标识号为首次接收到，则将所述人脸抓拍图像与所述服务器中人脸数据库进行比对，其中，所述服务器中人脸数据库中的每个人脸图像唯一对应一个第二标识号；所述服务器在所述人脸抓拍图像与所述服务器中人脸图像比对成功后，获取与所述人脸抓拍图像相似度值最高的所述服务器中人脸图像和对应的第二标识号； 所述服务器创建所述第一标识号与所述第二标识号的关联关系，所述关联关系用于关联所述摄像机中人脸数据库与所述服务器中人脸数据库。
本申请实施例网络摄像机采集到的图像由自身的处理器进行处理，并根据处理后的结果，有针对性的存储采集的图像，不仅降低了图像对比的难度，提高了对比结果的准确度，而且网络摄像机将图像处理后的结果发送给服务器，无需服务器进行图像对比，解决了由于网络摄像机采集的人脸图像无法在某些后端服务器上成功执行比对或比对结果准确率低的问题。
附图说明
为了更清楚地说明本申请实施例和现有技术的技术方案，下面对实施例和现有技术中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为本申请实施例提供的网络摄像机的结构示意图；
图2为图1所示实施例中网络摄像机将拍摄图像存储至第二存储单元的流程示意图；
图3为图1所示实施例中网络摄像机将拍摄图像上传至服务器的一种实现方式示意图；
图4为第二存储单元的属性设置界面图；
图5为第二存储单元中图像删除的一种实现方式示意图；
图6为显示设备展示的第二存储单元中目标图像的详细信息示意图；
图7为显示设备的人机交互界面的设计示意图；
图8为本申请实施提供的视频监控系统的结构示意图；
图9为本申请提供的更新摄像机中人脸数据库的方法实施例一的流程示意图；
图10为本申请提供的更新摄像机中人脸数据库的方法实施例二的流程示意图；
图11为本申请提供的摄像机和服务器中人脸数据库关联的方法实施例的流程示意图。
具体实施方式
为使本申请的目的、技术方案、及优点更加清楚明白，以下参照附图并举实施例，对本申请进一步详细说明。显然，所描述的实施例仅仅是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。下面以具体的实施例对本申请的技术方案进行详细说明。以下实施例均可以相互结合，对于相同、相似的概念或过程相互参见即可。
图1是本申请实施例提供的网络摄像机的结构示意图。如图1所示，本申请实施例提供的网络摄像机可以包括：图像传感器11、处理器12、存储器13和网络通信接口14。其中，该存储器13包括：第一存储单元131和第二存储单元132。
在本申请的实施例中，该图像传感器11，用于采集监控场景的图像；
该第一存储单元131，用于存储该图像传感器11采集到的所有图像，该第二存储单元132，用于存储图像传感器11采集到的部分图像，该部分图像为上述所有图像的子集。
其中，该处理器12，用于将图像传感器11采集的当前图像与该第二存储单元132中已存储的图像进行匹配，得到表示匹配结果的相似度值，将该相似度值与外界指示的第一相似度阈值和第二相似度阈值进行比较；在该相似度值小于第一相似度阈值但大于第二相似度阈值时，通过网络通信接口14以匹配失败消息将该当前图像发送给服务器，以及在该相似度值小于第一相似度阈值且小于第二相似度阈值时，通过该网络通信接口14以匹配失败消息将该当前图像发送给服务器，且为该当前图像分配一个唯一关联的设备目标标识，并将该当前图像存储至第二存储单元132。
其中，第一相似度阈值大于第二相似度阈值，该当前图像的设备目标标识是基于当前图像被存储至第二存储单元132的时间、网络摄像机的标志号以及随机值生成的。
外界指示的第一相似度阈值、第二相似度阈值，是指预先设置的第一相似度阈值、第二相似度阈值，当然第一相似度阈值、第二相似度阈值也可以 根据用户输入的指令进行调整。本领域技术人员可以理解的是，鉴于第一存储单元131容量的限制，第一存储单元131无法无限量的对图像进行存储，本申请实施例中的第一存储单元131用于存储该图像传感器11采集到的所有图像，是指图像传感器11采集到的所有图像均会送入第一存储单元131进行存储，当第一存储单元131存满后，可以对最早存入的图像进行覆盖，也可以更换新的第一存储单元131等。
可选的，匹配失败消息与下述的匹配成功消息可以通过不同协议的字段进行上传，以使服务器可以区别出网络摄像机上传时的匹配结果。
例如，可以在消息的头部中设置一个指定字段，用于区分匹配结果。匹配失败消息中该指定字段为第一字符(例如为0)，匹配成功消息中该指定字段为第二字符(例如为1)，可以通过识别该指定字段中的字符，来区分匹配结果。
可选的，在本实施例中，网络摄像机通过该图像传感器11采集其所处监控场景的图像，例如，当网络摄像机的镜头恰好定焦到一个人脸时，利用该图像传感器11抓拍包含该人脸的当前图像。一方面，图像传感器11将该当前图像传输至第一存储单元131进行存储，另一方面，该处理器12利用人脸检测技术，通过分析该当前图像中的人脸特征和第二存储单元132中已存储图像的人脸特征，将该当前图像与第二存储单元132中已存储的图像进行匹配，得到匹配结果，该匹配结果包括：该当前图像与上述已存储图像之间的相似度值。
其中，当前图像与已存储图像之间的相似度值可以包括当前图像与该第二存储单元132中所有已存储图像之间的所有相似度值，也可以为所有相似度值中的最大相似度值，本申请实施例并不对其进行限定，其可以根据实际情况自定义设置。
作为一种示例，处理器12首先将该相似度值与外界指示的第一相似度阈值进行匹配，确定出匹配结果。该第一相似度阈值用于判断当前图像与第二存储单元132中所有图像中的图像是否匹配成功。
在本实施例中，若当前图像与第二存储单元132中已存储的各图像的相似度值均小于第一相似度阈值，则表明第二存储单元132中不存在内容(本申请中的内容是指被关注的目标，例如，可以为人脸、人体或车辆等)与该当前 图像内容一致的图像，此时为了保证网络摄像机与服务器(即，监控视频系统中的后端服务器)中存储图像的同步，一方面可以利用网络通信接口14通过匹配失败消息将该当前图像发送给服务器，该匹配失败消息用于指示该当前图像在网络摄像机中匹配失败，即网络摄像机之前拍摄的图像中不存在内容一致的图像，以便服务器可以执行相应的处理。
另一方面，为了保证第二存储单元132中存储图像的唯一性，处理器12还可以将该相似度值与外界指示的第二相似度阈值进行匹配，以判定该当前图像是否满足存储至第二存储单元132的条件。该第二相似度阈值用于判断是否将当前图像存储至第二存储单元132中。在本实施例中，该第一相似度阈值大于第二相似度阈值，即匹配结果为匹配成功时，不会执行将当前图像存储至第二存储单元132的操作，因而，无需执行判断是否将该当前图像存入第二存储单元132，降低了运算量。
其中，在将该当前图像存储至第二存储单元132之前，可以首先为当前图像分配一个唯一关联的设备目标标识，该当前图像的设备目标标识可以根据当前图像被存储至第二存储单元132的时间、网络摄像机的标志号以及随机值生成，从而不仅能够保证生成的设备目标标识是唯一的，而且还可以标识出该当前图像是在何时由哪个网络摄像机拍摄的。
在本实施例中，图像的设备目标标识可以为用于表示该图像在设备内部的唯一标识，其可以采用64位字符，具体的组成方式可以为设备序列号+存储时间+随机数的方式。本申请并不对设备目标标识的具体组成方式进行限定，其可以根据实际情况确定。
示例性的，图2为图1所示实施例中网络摄像机将拍摄图像存储至第二存储单元的流程示意图。如图2所示，图像的存储过程包括如下步骤：S21、图像传感器采集到当前图像；S22、处理器判断该当前图像与第二存储单元中的图像的相似度值是否小于外界指示的第一相似度阈值；若否，执行S23、通过网络通信接口以匹配成功消息将该当前图像发送给服务器；若是，执行S24、通过网络通信接口以匹配失败消息将该当前图像发送给服务器，S25、判断该相似度值是否小于外界指示的第二相似度阈值，若是，执行S26、为当前图像分配一个唯一关联的设备目标标识，并将当前图像存储 至第二存储单元，若否，执行S27、根据外界指示进行处理。
其中，在S22中当前图像与第二存储单元中的图像的相似度值，具体可以为：当前图像与第二存储单元中各图像的最大相似度值。本申请实施例并不限定上述S27的具体实现原理，其可以基于实际情况确定。
在本申请的实施例中，网络摄像机利用图像传感器采集监控场景的图像，一方面，将采集到的图像存储至存储器的第一存储单元中，另一方面，利用处理器将图像传感器采集的当前图像与存储器的第二存储单元中已存储的图像进行匹配，得到表示匹配结果的相似度值，在相似度值小于外界指示的第一相似度阈值但大于外界指示的第二相似度阈值时，通过网络通信接口以匹配失败消息将当前图像发送给服务器，以及在相似度值小于第一相似度阈值且小于第二相似度阈值时，通过网络通信接口以匹配失败消息将当前图像发送给服务器的同时，还为当前图像分配一个唯一关联的设备目标标识，并将当前图像存储至第二存储单元中。
示例性的，在本实施例的一种可能设计中，上述处理器12具体用于在确定相似度值小于第一相似度阈值、且小于第二相似度阈值时，获取该当前图像的图像质量，在图像质量大于预设的图像质量时，通过网络通信接口14以匹配失败消息将该当前图像发送给服务器，且为该当前图像分配一个唯一关联的设备目标标识，并将当前图像存储至第二存储单元132。
在一些情况下，当图像质量较差时，可能会将不同内容的图像误判成相同的图像，因而，为了保证第二存储单元132中存储图像的质量，避免后续相似度判断时出现出错的问题，可选的，网络摄像机的处理器12在将当前图像存储至第二存储单元132之前，可以先做质量判断，在当前图像的图像质量大于预设的图像质量时，才将其存储至第二存储单元132，否则不对其进行存储。
在本申请实施例中，不对预设的图像质量的具体取值进行限定，其可以根据实际情况自定义设置，例如，可以根据第二存储单元132的存储空间大小，图像传感器11采集图像的最佳质量等进行确定，此处不再赘述。
本申请实施例中，通过对当前图像进行图像质量判断，可以提高第二存储单元中存储图像的质量，利用质量高的图像进行相似度判断，其结果的准确度也更高，为后续作出准确的相似度判断奠定基础。
进一步的，在本申请的一种实施例中，该处理器12还用于将相似度值与外界指示的第三相似度阈值进行比较，在相似度值大于第三相似度阈值时，获取当前图像的图像质量，若该图像质量大于当前图像对应匹配图像的图像质量时，利用该当前图像替换第二存储单元132中的匹配图像，并将该匹配图像的设备目标标识作为该当前图像的设备目标标识，其中，第三相似度阈值大于第一相似度阈值。
外界指示的第三相似度阈值是指预先设置的第三相似度阈值，当然第三相似度阈值也可以根据用户输入的指令进行调整。
在本实施例中，在本申请实施例中，处理器12在第一次存储图像时可以将所有满足第二相似度阈值以及预设的图像质量的图像均存储至第二存储单元132中，随着时间的推移，当网络摄像机处于光线较好的场景下时，拍摄到的当前图像的图像质量可能大于原有存入第二存储单元132的匹配图像，此时若当前图像与第二存储单元132中的匹配图像的相似度值大于第三相似度阈值，为了保证后续迅速、准确的确定出图像匹配结果，可以利用该当前图像替换第二存储单元132中的匹配图像。
可选的，为了保证上传到服务器的图像的设备目标标识是一致的，避免服务器重复处理，这时将该匹配图像的设备目标标识作为该当前图像的设备目标标识。
可选的，第三相似度阈值应当大于第一相似度阈值，示例性的，在本实施例中，当上述第一相似度阈值为85％时，该第三相似度阈值可以为88％、90％等取值，本申请实施例并不对第一相似度阈值和第三相似度阈值的具体取值进行限定，只要满足第三相似度阈值大于第一相似度阈值的条件均可。
本申请实施例的网络摄像机，第二存储单元中的图像具有更新机制，可以进一步降低网络摄像机的处理压力，提高了图像对比结果的准确度。
可选的，在本申请的另一种可能设计中，上述处理器12还用于确定第二存储单元132的当前存储量和第二存储单元132中图像的第一比例，根据当前存储量、第一比例和外界指示的第二存储单132的图像存储上限值，调整该第二相似度阈值，该第一比例用于指示当前存储量对应的目标对象数量与当前存储量的比值。
在本实施例中，第一相似度阈值和第二相似度阈值均可以是自定义的值，其中，该第二相似度阈值可以关联于第二存储单元132的存储量。具体的，处理器12可以基于第二存储单元132的存储量、第二存储单元132中图像的第一比例和外界指示的第二存储单元132的图像存储上限值，动态调整第二相似度阈值，使得该第二存储单元132内存储的图像数量不超过存储上限值。
其中，图像的第一比例也可以通过图像的存储重复率、图像的准确率等来表征，第一比例高，表征存储的准确率高，但表征存储的重复率低。具体的，第一比例表示第二存储单元132中存储的各图像中，不同目标的总数量与图像总数量的比。例如，第二存储单元132中共存储有100张人的图像，但该100张图像实际上只对应90个不同的人，那么第一比例为90％，也即，第二存储单元132中存储的图像的重复率为10％。外界指示的第二存储单元132的图像存储上限值具体可以为第二存储单元132中存储的不同目标的数量。可选的，第二存储单元132的存储量越大，第二相似度阈值越高；第二存储单元132中图像的第一比例越高，第二相似度阈值越高；外界指示的第二存储单元132的图像存储上限值越高，第二相似度阈值越高。
可选的，在本实施例中，处理器12还用于在当前存储量等于图像存储上限值时，利用当前图像替换第二存储单元中匹配成功次数最少的图像。
在当前图像有被存储至第二存储单元132的需求，但第二存储单元132已存满时，可以通过分析第二存储单元132中每张图像的匹配成功次数，确定出匹配成功次数最低的图像，该匹配成功次数最低的图像可能是被误存储至第二存储单元132的图像，或该图像对应的目标在该网络摄像机所监控的区域不常出现，因而，在第二存储单元132的当前存储量等于图像存储上限值时，为了保证当前图像能够被及时存储，可以利用当前图像替换第二存储单元132中匹配成功次数最少的图像。
示例性的，在本实施例的再一种可能的实施方式中，该处理器12还用于在相似度值大于或等于第一相似度阈值时，确定当前时间为本次匹配成功时间，判断该本次匹配成功时间与最近一次匹配成功时间之间的时间差是否大于或等于预设时间差，若是，通过网络通信接口14以匹配成功消息将当前图像发送给该服务器，并将最近一次匹配成功时间更新为本次匹配成功时间， 且对该当前图像对应匹配图像的匹配成功次数加1，若否，舍弃当前图像。
可选的，在本申请的实施例中，若该相似度值大于或等于第一相似度阈值，则表明第二存储单元132中存在内容与该当前图像内容一致的图像，这时为了避免内容的重复上传，判断目标匹配图像本次匹配成功时间与上一次匹配成功时间之间的时间差是否满足预设的时间关系，其中，目标匹配图像为第二存储单元132中与该当前图像相似度值最大的图像。
作为一种示例，若目标匹配图像本次匹配成功时间与上一次匹配成功时间之间的时间差大于或等于预设时间差，则表明图像传感器11没有在预设的时间段内重复采样，此时，一方面处理器12可以将该当前图像上传至服务器，具体的，可以通过网络通信接口14以匹配成功消息发送给该服务器，该匹配成功消息用于指示该当前图像在网络摄像机中匹配成功，即网络摄像机之前拍摄的图像中存在内容一致的图像，其可以表明网络摄像机之前已经上传过该当前图像；另一方面在网络摄像机中处理器12还可以对第二存储单元132中当前图像对应匹配图像的匹配成功次数加1，从而二存储单元132中匹配图像的匹配成功次数的统计。其中，当前图像匹配对应的匹配图像是指当前图像匹配成功的匹配图像，当前图像匹配对应的匹配图像的配成功次数越多，越能表明该当前图像的内容是经常采集到的内容。
作为另一种示例，若本次匹配成功时间与上一次匹配成功时间之间的时间差小于预设时间差，则表明图像传感器11在预设的时间段内重复采样，这时可以不对当前图像进行处理，或者直接将其舍弃。
对于预设时间差及其精度的具体取值可以根据实际情况确定，示例性的，在本实施例中，上述预设时间差可以为5s、8s、10s等不同的取值，预设时间差的精度可以为0.1s等。
示例性的，图3为图1所示实施例中网络摄像机将拍摄图像上传至服务器的一种实现方式示意图。如图3所示，该实现方式可以包括如下步骤：S31、图像传感器采集到当前图像；S32、处理器判断该当前图像与第二存储单元中的图像的相似度值是否小于外界指示的第一相似度阈值；若是，执行S33、通过网络通信接口以匹配失败消息将该当前图像发送给服务器；若否，执行S34、确定当前时间为本次匹配成功时间，S35、判断本次匹配成功 时间与最近一次匹配成功时间之间的时间差是否大于或等于预设时间差，若否，执行S36、舍弃当前图像；若是，执行S37、通过网络通信接口以匹配成功消息将当前图像发送给服务器，S38、将最近一次匹配成功时间更新为本次匹配成功时间，且对当前图像对应匹配图像的匹配成功次数加1。
其中，在S32中当前图像与第二存储单元中的图像的相似度值，具体可以为当前图像与第二存储单元中各图像的最大相似度值。
在本实施例中，在相似度值大于或等于第一相似度阈值时，通过将本次匹配成功时间与最近一次匹配成功时间之间的时间差和预设时间差进行对比，可以避免图像重复上传服务器，降低了服务器的处理负担，提高了视频监控系统的稳定性。
可选的，在上述实施例的基础中，上述处理器12还用于周期性获取第二存储单元132中每个图像在预设时间段内的匹配成功次数，保留匹配成功次数大于或等于外界指示的匹配成功次数阈值的图像，删除匹配成功次数小于外界指示的匹配成功次数阈值的图像。
外界指示的匹配成功次数阈值可以为预先设置的数值，也可以用户设置的数值。外界指示的匹配成功次数阈值与预设时间段正相关，例如预设时间段为1天时，匹配成功次数阈值可以为1次、2次或3次等；预设时间段为1周时，匹配成功次数阈值可以为5次、10次或20次等。
在本实施例中，网络摄像机还具有定期删除第二存储单元132中图像的功能，以保证第二存储单元132中存储的图像是经常使用的，从而提高第二存储单元132存储的图像的使用频率。
具体的，图像的添加、删除、以及针对第二存储单元132的属性设置等功能可以和通常情况下存储单元的管理完全兼容，其可以通过人机交互界面接收用户指定的属性值。
例如，图4为第二存储单元的属性设置界面图。如图4所示，在本实施例中，该第二存储单元的信息具体可以如下：
名称：第二存储单元1，第一相似度阈值(也即，匹配成功阈值)：85％，备注信息：无；
图像存储设置：
开启入库：未开启(其中，方框未选中表示未开启，方框选中表示开启)，第二相似度阈值(即，入库相似度阈值)：70％，图像质量阈值：60％；
清理设置：
判断周期：天，统计周期：2月，最小匹配成功次数：2次。
本申请实施例并不限定上述各参数的具体取值，其可以根据需求设定，上述给出的只是一种示例。
示例性的，图5为第二存储单元中图像删除的一种实现方式示意图。如图5所示，该实现方式可以包括如下步骤：S51、开启自动删除；S52、处理器周期性获取第二存储单元中每个图像在预设时间段内的匹配成功次数；S53、对于第二存储单元中每个图像，判断该图像的匹配成功次数是否大于或等于外界指示的匹配成功次数阈值；若是，执行S54、保留该图像；若否，执行S55、删除该图像。
在本实施例中，通过定期删除第二存储单元中匹配成功次数小于外界指示的匹配成功次数阈值的图像，可以保证第二存储单元中有足够的空间存储新添加的图像，从而减少因第二存储单元存满而导致异常的情况，提高了网络摄像机的自动化能力，提高了设备竞争力。
示例性的，在本申请实施例的又一种可能的设计中，上述处理器12还用于响应于外界的图像排序指示，根据该图像排序指示对第二存储单元132中的图像进行排序，得到排序结果，该图像排序指示用于指示第二存储单元132中的图像的排序方式，该排序方式依据的参数可以为：匹配成功次数、匹配成功时间、存储时间等，可以根据参数的大小采用由大到小或由小到大的顺序进行排序。
在本实施例中，该处理器12可以基于外界的图像排序指示对第二存储单元132中的图像进行排序，得到排序结果，这样处理器12可以将该排序结果传输至与该网络摄像机连接的显示设备，以使该显示设备呈现上述排序结果。
示例性的，该显示设备可以具有人机交互界面，该人机交互界面是可以被用户操控的，即用户可以通过该人机交互界面输入或生成图像排序指示，以指示第二存储单元132中的图像可以以何种排序方式进行展示。
示例性的，由于第二存储单元132中每个图像的设备目标标识均是由图像 被存储至第二存储单元132的时间、网络摄像机的标志号以及随机值生成的，且处理器12对每个图像的匹配成功次数、每次匹配成功的时间等信息进行了记录，所以，该网络摄像机至少可以提供匹配成功次数、匹配成功时间、存储时间等中的任意一种参数，用于对各图像进行排序。
示例性的，在本实施例中，上述处理器12还用于响应于外界发出的针对目标图像的展示指示，基于该展示指示，从第一存储单元131中获取与该目标图像之间的相似度大于第一相似度阈值的多张相似图像，以对该目标图像和多张相似图像进行展示。
在本实施例中，处理器12可以对图像传感器11拍摄到的所有图像进行存储，由于第二存储单元132中存储的图像为第一存储单元131中存储的图像的子集，具体的，第一存储单元131中存储的是拍摄到的所有图像，第二存储单元132中存储的是上述所有图像中满足第二相似度阈值的图像，因而，第一存储单元131中的图像与第二存储单元132中的图像具有一定的对应关系，例如，第一存储单元131中的图像可以携带匹配图像的标识，且具有匹配成功的时间，该匹配图像为第二存储单元132中的图像，因而，当用户发出展示指示时，处理器12可以响应于该展示指示，并基于展示指示中的目标图像的标识及上述对应关系，从第一存储空间中获取预设数量的相似图像，以实现对目标图像和多张相似图像的展示。
示例性的，网络摄像机可以与显示设备连接，因而，该显示设备可以显示网络摄像机中第一存储单元和/或第二存储单元中的图像。
可选的，显示设备具有人机交互界面，这样显示设备的人机交互界面呈现第二存储单元132中的图像时，用户可以通过该人机交互界面点击第二存储单元132中的某个图像，该人机交互界面上可以呈现出该图像的详细信息。图6为显示设备展示的第二存储单元中目标图像的详细信息示意图。如图6所示，显示设备的人机交互界面可以包括预览、回收(即，删除)、图片、应用、配置等功能选项，例如，对于应用选项，在第二存储单元展示界面上，具有筛选条件等，对于选中的一个图像，该人机交互界面还可以呈现该图像的详细信息，即图像详情。例如，图像属性，可以包括图像对应人员的姓名、性别、省份、城市、第二相似度阈值、存储时间、匹配成功总次数等信息，拍 摄记录可以包括：匹配时间、相识度、图像质量等。
示例性的，图7为显示设备的人机交互界面的设计示意图。如图7所示，在本实施例中，该人机交互界面可以划分为多个区域，例如，预览区域、匹配结果、抓拍展示、分析区域、第二存储单元、已删除图像展示等。关于各区域的具体分布详见图7所示，此处不再赘述。
本申请实施例提供的网络摄像机可以对在所处监控场景拍摄的图像进行处理，自动维护第二存储单元中的图像，减少了反复处理的过程，提高了网络摄像机的自动化程度，提高了图像对比结果的准确率，同时，网络摄像机可以为生成的每个图像分配一个唯一关联的设备目标标识，并且传输至服务器的图像也携带有设备目标标识，这样无需服务器进行图像匹配也可得到准确的图像对比结果。
进一步的，本申请还提供了一种视频监控系统。图8为本申请实施提供的视频监控系统的结构示意图。如图8所示，该视频监控系统可以包括：至少一个网络摄像机(例如，网络摄像机81至网络摄像机8n，其中，n为正整数)和服务器80，每个网络摄像机和服务器80通过网络建立通信连接。
上述至少一个网络摄像机中的各网络摄像机可以具有相同的配置，每个网络摄像机的实现原理类似，本实施例以上述至少一个网络摄像机中的一个网络摄像机81进行解释说明。
如图8所示，该网络摄像机81包括：图像传感器811、第一处理器812、第一存储器813和第一网络通信接口814，该第一存储器813包括：第一存储单元8131和第二存储单元8132。
其中，该图像传感器811，用于采集监控场景的图像；
该第一存储单元8131，用于存储所述图像传感器811采集到的所有图像，该第二存储单元8132，用于存储所述图像传感器811采集到的部分图像，该部分图像为所有图像的子集；
该第一处理器812，用于执行：
将所述图像传感器811采集到的当前图像与所述第二存储单元8132中已存储的图像进行匹配，得到表示匹配结果的相似度值，将所述相似度值与外界指示的第一相似度阈值、第二相似度阈值进行比较，在所述相似度值小于 第一相似度阈值但大于第二相似度阈值时，通过第一网络通信接口814以匹配失败消息将所述当前图像发送给服务器80，以及在相似度值小于第一相似度阈值且小于第二相似度阈值时，通过第一网络通信接口814以匹配失败消息将当前图像发送给服务器80，且为当前图像分配一个唯一关联的设备目标标识，并将所述当前图像存储至第二存储单元8132。
其中，所述第一相似度阈值大于所述第二相似度阈值，所述当前图像的设备目标标识是基于所述当前图像被存储至所述第二存储单元8132的时间、网络摄像机81的标志号以及随机值生成的。
关于该网络摄像机81以及其他网络摄像机的具体组成以及各部件的具体实现原理，可以参见上述图1至图7所示的记载，此处不再赘述。
如图8所示，该服务器80包括：第二网络通信接口821、第二处理器822和第二存储器823。
其中，该第二处理器822，用于通过第二网络通信接口821接收网络摄像机发送的匹配失败消息，判断第二存储器823中是否存在与当前图像的设备目标标识一致的图像，若否，为当前图像分配一个唯一关联的平台目标标识，并建立该设备目标标识与匹配失败消息的对应关系，以及存储该平台目标标识和该对应关系；若是，获取该当前图像在所述第二存储器823中的平台目标标识，建立所述设备目标标识与所述平台目标标识的映射关系，将所述当前图像和所述映射关系存储至所述第二存储器823。
在本实施例中，服务器80可以通过第二网络通信接口821接收多个网络摄像机发送的图像，本实施例以接收上述网络摄像机81发送的图像进行解释说明。
示例性的，当网络摄像机81的第一处理器812通过第一网络通信接口814以匹配失败消息向服务器80发送图像传感器811采集到的当前图像时，相应的，服务器80可以通过第二网络通信接口821接收该当前图像。
在本实施例中，由于服务器80接收到的当前图像是以匹配失败消息发送的，那么该服务器80的第二处理器822可以确定该网络摄像机之前可能没有发送过该当前图像的匹配图像，但是为了避免当前图像的匹配图像被遗漏记录，第二处理器822可以对该当前图像进行相似度匹配，判断该服务器80中是否存 储有该当前图像的匹配图像，也即，判断第二存储器823是否存在与当前图像的设备目标标识一致的图像，并根据判断结果执行相应的处理。
作为一种示例，若第二存储器823中不存在与当前图像的设备目标标识一致的图像，则建立该设备目标标识与匹配失败消息的对应关系，以记载该当前图像的内容未记载在服务器80中，从而做出标示，以便相关人员在后续进行处理。
作为另一种示例，若第二存储器823存在与当前图像的设备目标标识一致的图像，则获取该当前图像对应的匹配图像在第二存储器823中的平台目标标识，建立该设备目标标识与平台目标标识的映射关系，将并且将该映射关系存储至第二存储器823。
在本实施例中，虽然第二存储器823中存在该当前图像的匹配图像，但是其并没有记载该匹配图像与该当前图像之间的对应关系，因而，第二处理器822可以获取该匹配图像在第二存储器823中的平台目标标识，建立并存储该当前图像的设备目标标识与平台目标标识之间的映射关系，以便网络摄像机再次上传该当前图像的匹配图像时，可以直接获取两者之间的映射关系，以直接映射到前次的分析结果，减少了服务器80的算力占用，降低了服务器80的处理负担。
可选的，本实施例中的平台目标标识可以是该图像的内容在服务器中的详细信息，例如，人脸图像中人脸对应人员的实际身份信息，例如，可以为ID或名称等身份标识信息。本申请并不对平台目标标识的具有内容进行限定，其可以根据实际需求设定。
示例性的，在本申请的一种可能的设计中，若第一处理器812，还用于在相似度值大于或等于第一相似度阈值时，通过该第一网络通信接口814以匹配成功消息将当前图像发送给所述服务器80；
相应的，第二处理器822，还用于通过第二网络通信接口821接收网络摄像机发送的匹配成功消息，获取该当前图像对应匹配图像在第二存储器823中的平台目标标识，将该平台目标标识的分析结果作为当前图像的分析结果。
在本实施例中，若服务器80接收到的当前图像是以匹配成功消息发送的，那么该服务器80的第二处理器822可以确定该网络摄像机之前可能发送过该 当前图像的匹配图像，则直接获取该当前图像对应匹配图像在第二存储器823中的平台目标标识，将该平台目标标识的分析结果作为当前图像的分析结果，无需再次对当前图像进行分析，减少了服务器80端的算力占用，减少了服务器80的处理任务，降低了服务器80的处理负担。
进一步的，在上述实施例的基础上，第二处理器822，还用于通过第二网络通信接口821将该当前图像的平台目标标识发送给网络摄像机；
相应的，第一处理器812还用于通过第一网络通信接口814接收该当前图像的平台目标标识，并在该图像传感器811获取到的图像与第二存储单元8132中已存储的图像的相似度值大于或等于第一相似度阈值时，将该平台目标标识作为图像的设备目标标识发送给服务器80；
相应的，该第二处理器822，还用于直接根据接收到的图像的平台目标标识，获取该图像的分析结果。
在本实施例中，当第一处理器812以匹配成功消息上传当前图像时，表明网络摄像机的第二存储单元8132中存储有该当前图像的匹配图像，这时第二处理器822可以将获取到与该当前图像的平台目标标识发送给网络摄像机。
网络摄像机接收到该平台目标标识后，一方面可以将平台目标标识直接作为当前图像对应匹配图像的设备目标标识，这样网络摄像机再次上传匹配成功的图像时可以以该平台目标标识作为设备目标标识发送给服务器80，这样服务器80的第二处理器822可以根据该平台目标标识直接获取图像的分析结果，无需获取与设备目标标识具有映射关系的平台目标标识，减少了映射关系查找的工作量。
另一方面，网络摄像机可以保持将平台目标标识与设备目标标识的映射关系，当采集到的图像匹配成功时，可以首先确定出平台目标标识再将平台目标标识上传，同样可以减少服务器80的映射关系查找的工作量，减低了服务器80的处理负担。
本申请实施例提供的视频监控系统，网络摄像机可以对当前点位的第二存储单元中的图像进行自维护，针对当前点位通过长期的积累形成一个点位特有的第二存储单元，在后端服务器形成一个设备中图像的设备目标标识和 后端大型存储器中图像的平台目标标识的映射关系，后续接收的图像一旦匹配成功，则服务器不需要再次进行匹配，而是将网络摄像机中匹配成功的图像的设备目标标识直接进行映射，不仅减少了处理压力，同时规避了前端网络摄像机和服务器不兼容和第二存储单元中图像维护困难的问题，
示例性的，在本申请的另一种示例性中，图9为本申请提供的更新摄像机中人脸数据库的方法实施例一的流程示意图。如图9所示，在本实施例中，该方法可以包括如下步骤：
S91：响应于一个人脸抓拍指令，获取一个人脸抓拍图像。
在本实施例中，摄像机可以在外界的触发作用下执行人脸抓拍图像采集操作。示例性的，该摄像机可以获取外界指示的人脸抓拍指令，根据该人脸抓拍指令，利用摄像机的图像传感器采集人脸抓拍图像，相应的，摄像机的处理器对获取到的至少一个人脸抓拍图像进行处理。
S92：根据摄像机本地存储的人脸数据库，匹配该人脸抓拍图像和人脸数据库中任一人脸图像，计算该人脸抓拍图像的相似度值。
其中，该人脸数据库中存储着至少两个人脸图像，且每个人脸图像唯一对应着一个标识号；其中，该标识号用于指示其对应的人脸图像存储至人脸数据库的时间；且标识号对应着一个频次数值，该频次数值用于指示标识号对应的人脸图像发生匹配成功的次数。
示例性的，该标识号可以包括摄像机的序列标识号、随机值和对应的人脸图像存储至该人脸数据库的时间。
在本实施例中，摄像机本地维护有一个人脸数据库，该人脸数据库中存储了该摄像机抓拍到的部分图像。通常情况下，该人脸数据库中的至少两个人脸图像通过采集不同用户的人脸得到。具体的，每个人脸图像唯一对应着一个标识号、一个频次数值。
因而，当摄像机获取到一个新的人脸抓拍图像时，可以将该人脸抓拍图像与本地的人脸数据库中的人脸图像进行匹配，确定出人脸抓拍图像的相似度值，以确定是否需要对该人脸数据库进行更新。
S93：在该人脸抓拍图像的相似度值大于第一阈值时，确定匹配成功，且获取该人脸数据库中与人脸抓拍图像相似度值最高的人脸图像、该人脸图像 对应的第一标识号以及该第一标识号对应的频次数值。
在本实施例中，当确定人脸抓拍图像的相似度值大于第一阈值时，表明该人脸数据库中存在与该人脸抓拍图像内容基本一致的人脸图像，因而，为了避免相同内容的图像被重复存储，则需要确定出该人脸数据库中与该人脸抓拍图像相似度值最高的人脸图像，相应的，确定出该相似度值最高的人脸图像对应的第一标识号以及该第一标识号对应的频次数值。
进一步的，在本申请的实施例中，在获取第一标识号后，该方法还包括：将该人脸抓拍图像和第一标识号发送至服务器。
通过将人脸抓拍图像和第一标识号共同发送至服务器，使得服务器可以根据该标识号确定出之前对该人脸抓拍图像对应匹配图像的处理结果，简化了服务器的处理操作。
S94：在该第一标识号对应的频次数值小于预设阈值时，删除摄像机中人脸数据库中第一标识号对应的人脸图像，更新该摄像机中人脸数据库。
在本实施例中，频次数值对应的预设阈值可以用于指示人脸图像是否允许被存储在人脸数据库的判定条件。当第一标识号对应的频次数值小于预设阈值时，表明该人脸图像可能是被误存储至人脸数据库的图像或该图像对应的目标在该摄像机所处区域不常出现，这时可以将摄像机中人脸数据库中第一标识号对应的人脸图像删除，更新该摄像机中人脸数据库，从而保证该人脸数据库中存储图像的高准确率。
本领域技术人员可以理解的是，步骤S94描述的执行过程，是摄像机在运行一段时间后才触发的，或针对的为人脸数据库的剩余存储空间不足，即小于设定的容量阈值的场景。
可选的，S94：在该第一标识号对应的频次数值小于预设阈值时，删除摄像机中人脸数据库中第一标识号对应的人脸图像，更新该摄像机中人脸数据库，包括：在所述第一标识号对应的人脸图像存储的时长大于预设时长和/或所述摄像机中人脸数据库的剩余存储空间小于预设容量阈值的情况下，当该第一标识号对应的频次数值小于预设阈值时，删除摄像机中人脸数据库中第一标识号对应的人脸图像，更新该摄像机中人脸数据库。
可选的，在本实施例的一种可能设计中，在S94之前，该方法还可以先执 行如下操作，再执行S94。也即，在S93之后，对上述第一标识号对应的频次数值做加1运算，并记录运算后的频次数值，从而实现对第一标识号对应频次数值的更新。
相应的，该S94可以替换为如下步骤：
在运算后的频次数值小于预设阈值时，删除该摄像机中人脸数据库中第一标识号对应的人脸图像。
将运算后的频次数值与预设阈值进行比较，有利于提高判定人类数据库中存储图像是否准确的准确度。
本申请实施例提供的方法，通过响应于一个人脸抓拍指令，获取一个人脸抓拍图像，根据摄像机本地存储的人脸数据库，匹配人脸抓拍图像和人脸数据库中任一人脸图像，计算人脸抓拍图像的相似度值；在人脸抓拍图像的相似度值大于第一阈值时，确定匹配成功，且获取人脸数据库中与人脸抓拍图像相似度值最高的人脸图像、人脸图像对应的第一标识号以及第一标识号对应的频次数值；在第一标识号对应的频次数值小于预设阈值时，删除摄像机中人脸数据库中第一标识号对应的人脸图像，更新该摄像机中人脸数据库，保证了人脸数据库中存储图像的准确度，提高了图像对比结果的准确率。
示例性的，在上述实施例的基础上，图10为本申请提供的更新摄像机中人脸数据库的方法实施例二的流程示意图。如图10所示，在本实施例中，该方法还可以包括如下步骤：
S101：在人脸抓拍图像的相似度值小于第一阈值时，判断人脸抓拍图像的相似度值是否小于第二阈值，若是，执行步骤S102，若否，执行步骤S103。
可以理解的是，该S101可以位于上述S92之后执行，也即，在人脸抓拍图像的相似度值小于第一阈值时，再比较人脸抓拍图像的相似度值和第二阈值的大小，在根据对比结果执行相应的操作。
S102：将该人脸抓拍图像存储至摄像机中人脸数据库，记录该人脸抓拍图像存储至摄像机中人脸数据库的时间，为该人脸抓拍图像分配一个唯一标识号。
其中，该第二阈值小于或等于第一阈值。
在本实施例中，在人脸抓拍图像的相似度值小于第二阈值时，表明人脸 数据库中不存在内容与该人脸抓拍图像内容一致的图像，这时可以将人脸抓拍图像存储至摄像机中人脸数据库。此外，为了便于后续对该人脸数据库的维护，还可以记录该人脸抓拍图像存储至摄像机中人脸数据库的时间，并为该人脸抓拍图像分配一个唯一标识号，用以唯一标识该人脸抓拍图像。
S103：不将所述人脸抓拍图像存储至所述摄像机中人脸数据库，但将该人脸抓拍图像发送至服务器。
在本实施例中，在人脸抓拍图像的相似度值大于或等于第二阈值时，表明人脸数据库中已经存在内容与该人脸抓拍图像内容一致的图像，为了避免人脸图像的重复存储，则不将人脸抓拍图像存储至摄像机中人脸数据库，但将会人脸抓拍图像发送至服务器，以使服务器对该人脸抓拍图像执行相应的处理。
本申请实施例的方法，在人脸抓拍图像的相似度值小于第一阈值时，比较人脸抓拍图像的相似度值和第二阈值，在人脸抓拍图像的相似度值小于第二阈值时，将人脸抓拍图像存储至摄像机中人脸数据库，记录人脸抓拍图像存储至摄像机中人脸数据库的时间，为该人脸抓拍图像分配一个唯一的标识号。
进一步的，在本申请的实施例中，该方法还可以包括如下步骤：
周期性获取摄像机中人脸数据库中所有人脸图像对应的频次数值；
对于小于预设频次数值的人脸图像进行删除，更新该摄像机中的人脸数据库。
进一步的，在本申请的实施例中，该方法还可以包括：响应于接收到的关于人脸数据库的显示指令，生成一个数据包，该数据包用于显示人脸数据库中与该显示指令对应的人脸图像。
在本实施例中，摄像机还可以获取外界发出的关于该人脸数据库的显示指令，基于该显示指令对人脸数据库中的图像进行排序、统计等处理，生成一个数据包，这样摄像机可以将该数据包传输至与该摄像机连接的显示设备，以使该显示设备显示人脸数据库中与该显示指令对应的人脸图像。
本申请实施例提供的更新摄像机中人脸数据库的方法，摄像机可以自动维护和更新本地的人脸数据库，减少了图像反复处理的过程，提高了摄像机 的自动化程度，提高了图像对比结果的准确率，同时，摄像机可以为每个人脸图像分配一个唯一对应的标识号，并且传输至服务器的图像也携带有标识号。
本实施例中的摄像机也即上述实施例中的网络摄像机，本申请中的人脸数据库也即上述实施例中存储器的第二存储单元，关于本实施例中未详尽的描述均可以参见上述实施例中的记载，此处不再赘述。
示例性的，在本申请的再一种示例性中，图11为本申请提供的摄像机和服务器中人脸数据库关联的方法实施例的流程示意图。如图11所示，在本实施例中，该方法可以包括如下步骤：
S111：摄像机响应于一个人脸抓拍指令，获取一个人脸抓拍图像。
S112：摄像机根据摄像机本地存储的人脸数据库，匹配该人脸抓拍图像和人脸数据库中任一人脸图像，计算该人脸抓拍图像的相似度值。
其中，该人脸数据库中存储着至少两个人脸图像，且每个人脸图像唯一对应着一个标识号；其中，该标识号用于指示其对应的人脸图像存储至摄像机中人脸数据库的时间；可选的，标识号对应着一个频次数值，该频次数值用于指示标识号对应的人脸图像发生匹配成功的次数。
S113：摄像机在该人脸抓拍图像的相似度值大于第一阈值时，确定匹配成功，且获取该人脸数据库中与所述人脸抓拍图像相似度值最高的人脸图像和所述人脸图像对应的第一标识号；
S114：摄像机将第一标识号和人脸抓拍图像发送至服务器；
S115：服务器根据接收到的所述第一标识号，响应于该第一标识号为首次接收到，则将该人脸抓拍图像与服务器中人脸数据库进行比对。
其中，该服务器中人脸数据库中的每个人脸图像唯一对应一个第二标识号；
S116：服务器在该人脸抓拍图像与服务器中人脸图像比对成功后，获取与人脸抓拍图像相似度值最高的所述服务器中人脸图像和对应的第二标识号。
S117：服务器创建该第一标识号与第二标识号的关联关系，该关联关系用于关联该摄像机中人脸数据库与服务器中人脸数据库。
在一种可能的实施方式中，上述摄像机和服务器中人脸数据库关联的方 法，还包括：当所述人脸数据库的人脸图像存储的时长大于第一预设时长、且该人脸图像的标识号对应的匹配成功次数小于预设阈值时，在所述人脸数据库中删除该人脸图像。和/或当所述人脸数据库的剩余存储空间小于预设容量阈值的情况下，在所述人脸数据库中标识号对应的匹配成功次数小于预设阈值的人脸图像。
本申请的实施例中，摄像机获取一个人脸抓拍图像后，根据摄像机本地存储的人脸数据库，匹配该人脸抓拍图像和人脸数据库中任一人脸图像，计算该人脸抓拍图像的相似度值，在该人脸抓拍图像的相似度值大于第一阈值时，确定匹配成功并获取该人脸数据库中与人脸抓拍图像相似度值最高的人脸图像和人脸图像对应的第一标识号，以及将第一标识号和人脸抓拍图像发送至服务器，相应的，服务器根据接收到的第一标识号，响应于第一标识号为首次接收到，则将该人脸抓拍图像与所述服务器中人脸数据库进行比对，在该人脸抓拍图像与服务器中人脸图像比对成功后，获取与人脸抓拍图像相似度值最高的服务器中人脸图像和对应的第二标识号，创建该第一标识号与第二标识号的关联关系，该关联关系用于关联该摄像机中人脸数据库与服务器中人脸数据库。
可以理解的是，本实施例中某些步骤的具体实现可以参见上述任一实施例中的记载，关于本实施例中未详尽的描述均可以参见上述实施例中的记载，此处不再赘述。
最后应说明的是：以上各实施例仅用以说明本申请的技术方案，而非对其限制；尽管参照前述各实施例对本申请进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分或者全部技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本申请各实施例技术方案的范围。
以上所述仅为本申请的较佳实施例，并不用以限制本申请，凡在本申请的精神和原则之内，所做的任何修改、等同替换、改进等，均应包含在本申请保护的范围之内。

Claims (20)

  1. 一种网络摄像机，其特征在于，包括：图像传感器、处理器、存储器和网络通信接口；所述存储器包括：第一存储单元和第二存储单元；
    所述图像传感器，用于采集监控场景的图像；
    所述第一存储单元，用于存储所述图像传感器采集到的所有图像，所述第二存储单元，用于存储所述图像传感器采集到的部分图像，所述部分图像为所述所有图像的子集；
    所述处理器，用于执行：
    将所述图像传感器采集的当前图像与所述第二存储单元中已存储的图像进行匹配，得到表示匹配结果的相似度值，将所述相似度值与外界指示的第一相似度阈值、第二相似度阈值进行比较；
    在所述相似度值小于所述第一相似度阈值但大于所述第二相似度阈值时，通过所述网络通信接口以匹配失败消息将所述当前图像发送给服务器；
    在所述相似度值小于所述第一相似度阈值且小于所述第二相似度阈值时，通过所述网络通信接口以匹配失败消息将所述当前图像发送给服务器，且为所述当前图像分配一个唯一关联的设备目标标识，并将所述当前图像存储至所述第二存储单元；
    其中，所述第一相似度阈值大于所述第二相似度阈值，所述设备目标标识是基于所述当前图像被存储至所述第二存储单元的时间、所述网络摄像机的标志号以及随机值生成的。
  2. 根据权利要求1所述的网络摄像机，其特征在于，所述处理器具体用于在所述相似度值小于所述第一相似度阈值且小于所述第二相似度阈值时，获取所述当前图像的图像质量，在所述图像质量大于预设的图像质量时，通过所述网络通信接口以匹配失败消息将所述当前图像发送给服务器，且为所述当前图像分配一个唯一关联的设备目标标识，并将所述当前图像存储至所述第二存储单元。
  3. 根据权利要求1所述的网络摄像机，其特征在于，所述处理器还用于将所述相似度值与外界指示的第三相似度阈值进行比较，在所述相似度值大于所述第三相似度阈值时，获取所述当前图像的图像质量，若所述图像质量大 于所述当前图像对应匹配图像的图像质量时，利用所述当前图像替换所述第二存储单元中的所述匹配图像，并将所述匹配图像的设备目标标识作为所述当前图像的设备目标标识，所述第三相似度阈值大于所述第一相似度阈值。
  4. 根据权利要求1所述的网络摄像机，其特征在于，所述处理器还用于确定所述第二存储单元的当前存储量和所述第二存储单元中图像的第一比例，根据所述当前存储量、所述第一比例和外界指示的所述第二存储单元的图像存储上限值，调整所述第二相似度阈值，所述第一比例用于指示所述当前存储量对应的目标对象数量与所述当前存储量的比值；
    所述处理器还用于在所述当前存储量等于所述图像存储上限值时，利用所述当前图像替换所述第二存储单元中匹配成功次数最少的图像。
  5. 根据权利要求1所述的网络摄像机，其特征在于，所述处理器还用于在所述相似度值大于或等于所述第一相似度阈值时，确定当前时间为本次匹配成功时间，判断所述本次匹配成功时间与最近一次匹配成功时间之间的时间差是否大于或等于预设时间差，若是，通过所述网络通信接口以匹配成功消息将所述当前图像发送给所述服务器，并将所述最近一次匹配成功时间更新为所述本次匹配成功时间，且对所述当前图像对应匹配图像的匹配成功次数加1。
  6. 根据权利要求5所述的网络摄像机，其特征在于，所述处理器还用于周期性获取所述第二存储单元中每个图像在预设时间段内的匹配成功次数，保留匹配成功次数大于或等于外界指示的匹配成功次数阈值的图像，删除匹配成功次数小于外界指示的匹配成功次数阈值的图像。
  7. 根据权利要求1所述的网络摄像机，其特征在于，所述处理器还用于响应于外界的图像排序指示，根据所述图像排序指示对所述第二存储单元中的图像进行排序，得到排序结果，所述图像排序指示用于指示对所述第二存储单元中的图像的排序方式，所述排序方式包括如下任意一种：匹配成功次数、匹配成功时间、存储时间。
  8. 根据权利要求7所述的网络摄像机，其特征在于，所述处理器还用于响应于外界发出的针对目标图像的展示指示，基于所述展示指示，从所述第一存储单元中获取与所述目标图像之间的相似度大于所述第一相似度阈值的多 张相似图像，以对所述目标图像和所述多张相似图像进行展示。
  9. 一种视频监控系统，其特征在于，包括：至少一个网络摄像机和服务器，每个网络摄像机和所述服务器通过网络建立通信连接；
    对于所述至少一个网络设备中的任意一个网络摄像机，所述网络摄像机包括：图像传感器、第一处理器、第一存储器和第一网络通信接口，所述第一存储器包括：第一存储单元和第二存储单元；
    所述图像传感器，用于采集监控场景的图像；
    所述第一存储单元，用于存储所述图像传感器采集到的所有图像，所述第二存储单元，用于存储所述图像传感器采集到的部分图像，所述部分图像为所述所有图像的子集；
    所述第一处理器，用于执行：
    将所述图像传感器采集到的当前图像与所述第二存储单元中已存储的图像进行匹配，得到表示匹配结果的相似度值，将所述相似度值与外界指示的第一相似度阈值、第二相似度阈值进行比较，在所述相似度值小于所述第一相似度阈值但大于所述第二相似度阈值时，通过所述第一网络通信接口以匹配失败消息将所述当前图像发送给所述服务器，以及在所述相似度值小于所述第一相似度阈值且小于所述第二相似度阈值时，通过所述网络通信接口以匹配失败消息将所述当前图像发送给服务器，且为所述当前图像分配一个唯一关联的设备目标标识，并将所述当前图像存储至所述第二存储单元，其中，所述第一相似度阈值大于所述第二相似度阈值，所述设备目标标识是基于所述当前图像被存储至所述第二存储单元的时间、所述网络摄像机的标志号以及随机值生成的；
    其中，所述服务器包括：第二网络通信接口、第二处理器和第二存储器；
    所述第二处理器，用于通过所述第二网络通信接口接收所述网络摄像机发送的匹配失败消息，判断所述第二存储器是否存在与所述当前图像的设备目标标识一致的图像，若否，为所述当前图像分配一个唯一关联的平台目标标识，并建立所述设备目标标识与匹配失败的对应关系，以及存储所述平台目标标识和所述对应关系；若是，获取所述当前图像对应匹配图像在所述第二存储器中的平台目标标识，建立所述设备目标标识与所述平台目标标识的 映射关系，将所述映射关系存储至所述第二存储器。
  10. 一种更新摄像机中人脸数据库的方法，其特征在于，包括：
    响应于一个人脸抓拍指令，获取一个人脸抓拍图像；
    根据所述摄像机本地存储的人脸数据库，匹配所述人脸抓拍图像和所述人脸数据库中任一人脸图像，计算所述人脸抓拍图像的相似度值；其中，所述人脸数据库中存储着至少两个人脸图像，且每个人脸图像唯一对应着一个标识号；其中，所述标识号用于指示其对应的人脸图像存储至所述人脸数据库的时间；且所述标识号对应着一个频次数值，所述频次数值用于指示所述标识号对应的人脸图像发生匹配成功的次数；
    在所述人脸抓拍图像的相似度值大于第一阈值时，确定匹配成功，且获取所述人脸数据库中与所述人脸抓拍图像相似度值最高的人脸图像、所述人脸图像对应的第一标识号以及所述第一标识号对应的频次数值；
    在所述第一标识号对应的频次数值小于预设阈值时，删除所述摄像机中人脸数据库中所述第一标识号对应的人脸图像，更新所述摄像机中人脸数据库。
  11. 根据权利要求10所述的方法，其特征在于，所述方法还包括：
    在所述人脸抓拍图像的相似度值小于所述第一阈值时，比较所述人脸抓拍图像的相似度值和第二阈值；
    在所述人脸抓拍图像的相似度值小于所述第二阈值时，将所述人脸抓拍图像存储至所述摄像机中人脸数据库，记录所述人脸抓拍图像存储至所述摄像机中人脸数据库的时间，为所述人脸抓拍图像分配一个唯一标识号；其中，所述第二阈值小于或等于所述第一阈值。
  12. 根据权利要求11所述的方法，其特征在于，所述方法还包括：
    在所述人脸抓拍图像的相似度值大于所述第二阈值时，不将所述人脸抓拍图像存储至所述摄像机中人脸数据库，但将所述人脸抓拍图像发送至服务器。
  13. 根据权利要求10所述的方法，其特征在于，所述标识号包括所述摄像机的序列标识号、随机值和对应的人脸图像存储至所述人脸数据库的时间。
  14. 根据权利要求13所述的方法，其特征在于，在所述第一标识号对应的 频次数值小于预设阈值时，删除所述摄像机中人脸数据库中所述第一标识号对应的人脸图像之前，包括：
    对所述第一标识号对应的频次数值做加1运算，并记录运算后的频次数值；
    相应的，所述在所述第一标识号对应的频次数值小于预设阈值时，删除所述摄像机中人脸数据库中所述第一标识号对应的人脸图像，包括：
    在所述运算后的频次数值小于预设阈值时，删除所述摄像机中人脸数据库中所述第一标识号对应的人脸图像。
  15. 根据权利要求14所述的方法，其特征在于，所述方法还包括：
    周期性获取所述摄像机中人脸数据库中所有人脸图像对应的频次数值；
    对于小于预设频次数值的人脸图像进行删除，更新所述摄像机中的所述人脸数据库。
  16. 根据权利要求15所述的方法，其特征在于，在获取所述第一标识号后，所述方法还包括：
    将所述人脸抓拍图像和所述第一标识号发送至服务器。
  17. 根据权利要求15所述的方法，其特征在于，所述方法还包括：
    响应于接收到的关于所述人脸数据库的显示指令，生成一个数据包，所述数据包用于显示所述人脸数据库中与所述显示指令对应的人脸图像。
  18. 一种摄像机和服务器中人脸数据库关联的方法，其特征在于，包括：
    所述摄像机响应于一个人脸抓拍指令，获取一个人脸抓拍图像；
    所述摄像机根据所述摄像机本地存储的人脸数据库，匹配所述人脸抓拍图像和所述人脸数据库中任一人脸图像，计算所述人脸抓拍图像的相似度值；其中，所述人脸数据库中存储着至少两个人脸图像，且每个人脸图像唯一对应着一个标识号；其中，所述标识号用于指示其对应的人脸图像存储至所述摄像机中人脸数据库的时间；
    所述摄像机在所述人脸抓拍图像的相似度值大于第一阈值时，确定匹配成功，且获取所述人脸数据库中与所述人脸抓拍图像相似度值最高的人脸图像和所述人脸图像对应的第一标识号；将所述第一标识号和所述人脸抓拍图像发送至所述服务器；
    所述服务器根据接收到的所述第一标识号，响应于所述第一标识号为首 次接收到，则将所述人脸抓拍图像与所述服务器中人脸数据库进行比对，其中，所述服务器中人脸数据库中的每个人脸图像唯一对应一个第二标识号；
    所述服务器在所述人脸抓拍图像与所述服务器中人脸图像比对成功后，获取与所述人脸抓拍图像相似度值最高的所述服务器中人脸图像和对应的第二标识号；
    所述服务器创建所述第一标识号与所述第二标识号的关联关系，所述关联关系用于关联所述摄像机中人脸数据库与所述服务器中人脸数据库。
  19. 根据权利要求18所述的方法，其特征在于，所述标识号对应着一个频次数值，所述频次数值用于指示所述标识号对应的人脸图像发生匹配成功的次数。
  20. 根据权利要求19所述的方法，其特征在于，所述方法还包括：
    当所述人脸数据库的人脸图像存储的时长大于第一预设时长、且该人脸图像的标识号对应的匹配成功次数小于预设阈值时，在所述人脸数据库中删除该人脸图像；
    和/或当所述人脸数据库的剩余存储空间小于预设容量阈值的情况下，在所述人脸数据库中标识号对应的匹配成功次数小于预设阈值的人脸图像。
PCT/CN2020/109328 2019-08-29 2020-08-14 网络摄像机、视频监控系统及方法 WO2021036832A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP20857136.4A EP4024275A4 (en) 2019-08-29 2020-08-14 NETWORK CAMERA, VIDEO SURVEILLANCE SYSTEM AND METHOD
US17/753,223 US11750776B2 (en) 2019-08-29 2020-08-14 Network camera, video monitoring system and method

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910808803.2A CN112446395B (zh) 2019-08-29 2019-08-29 网络摄像机、视频监控系统及方法
CN201910808803.2 2019-08-29

Publications (1)

Publication Number Publication Date
WO2021036832A1 true WO2021036832A1 (zh) 2021-03-04

Family

ID=74685075

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/109328 WO2021036832A1 (zh) 2019-08-29 2020-08-14 网络摄像机、视频监控系统及方法

Country Status (4)

Country Link
US (1) US11750776B2 (zh)
EP (1) EP4024275A4 (zh)
CN (1) CN112446395B (zh)
WO (1) WO2021036832A1 (zh)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113126929A (zh) * 2021-04-23 2021-07-16 重庆紫光华山智安科技有限公司 一种特征数据去重的方法、系统、介质和终端
CN113409056A (zh) * 2021-06-30 2021-09-17 深圳市商汤科技有限公司 支付方法、装置、本地识别设备、人脸支付系统及设备
CN114584617A (zh) * 2022-05-06 2022-06-03 南京维数软件股份有限公司 一种基于汇聚处理平台的抓拍数据分发方法及装置
CN115684688A (zh) * 2022-03-31 2023-02-03 江苏未来智慧信息科技有限公司 针对防窃电系统的图像监测传递系统与方法

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113449628A (zh) * 2021-06-24 2021-09-28 北京澎思科技有限公司 图像处理系统及方法、装置、存储介质、计算机程序产品
CN113596325B (zh) * 2021-07-15 2023-05-05 盛景智能科技(嘉兴)有限公司 抓图方法、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
US20180225506A1 (en) * 2017-02-08 2018-08-09 Safran Identity & Security Device for processing images of people, the device seeking to sort these images as a function of contextual information
CN108921150A (zh) * 2018-09-18 2018-11-30 深圳市华百安智能技术有限公司 基于网络硬盘录像机的人脸识别系统
CN109241897A (zh) * 2018-08-29 2019-01-18 深圳华远云联数据科技有限公司 监控图像的处理方法、装置、网关设备及存储介质
CN109800691A (zh) * 2019-01-07 2019-05-24 深圳英飞拓科技股份有限公司 基于人脸识别技术的人口统计方法及系统

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080080748A1 (en) 2006-09-28 2008-04-03 Kabushiki Kaisha Toshiba Person recognition apparatus and person recognition method
US20160088326A1 (en) * 2014-09-23 2016-03-24 Watchcorp Holdings LLC Distributed recording, managing, and accessing of surveillance data within a networked video surveillance system
US9747494B2 (en) * 2015-11-16 2017-08-29 MorphoTrak, LLC Facial matching system
CN106790054A (zh) * 2016-12-20 2017-05-31 四川长虹电器股份有限公司 基于人脸识别和声纹识别的交互式认证系统及方法
CN107590212A (zh) * 2017-08-29 2018-01-16 深圳英飞拓科技股份有限公司 一种人脸图片的入库系统及方法
CN108429782B (zh) * 2017-09-12 2020-11-06 腾讯科技（深圳）有限公司 信息推送方法、装置、终端及服务器
US9985786B1 (en) * 2017-11-21 2018-05-29 Ca, Inc. Cross-device authentication
CN109936522B (zh) * 2017-12-15 2022-03-04 株式会社日立制作所 设备认证方法及设备认证系统
CN109658572B (zh) 2018-12-21 2020-09-15 上海商汤智能科技有限公司 图像处理方法及装置、电子设备和存储介质
CN109858371B (zh) * 2018-12-29 2021-03-05 深圳云天励飞技术有限公司 人脸识别的方法及装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180225506A1 (en) * 2017-02-08 2018-08-09 Safran Identity & Security Device for processing images of people, the device seeking to sort these images as a function of contextual information
CN109241897A (zh) * 2018-08-29 2019-01-18 深圳华远云联数据科技有限公司 监控图像的处理方法、装置、网关设备及存储介质
CN108921150A (zh) * 2018-09-18 2018-11-30 深圳市华百安智能技术有限公司 基于网络硬盘录像机的人脸识别系统
CN109800691A (zh) * 2019-01-07 2019-05-24 深圳英飞拓科技股份有限公司 基于人脸识别技术的人口统计方法及系统

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113126929A (zh) * 2021-04-23 2021-07-16 重庆紫光华山智安科技有限公司 一种特征数据去重的方法、系统、介质和终端
CN113126929B (zh) * 2021-04-23 2022-04-22 重庆紫光华山智安科技有限公司 一种特征数据去重的方法、系统、介质和终端
CN113409056A (zh) * 2021-06-30 2021-09-17 深圳市商汤科技有限公司 支付方法、装置、本地识别设备、人脸支付系统及设备
CN115684688A (zh) * 2022-03-31 2023-02-03 江苏未来智慧信息科技有限公司 针对防窃电系统的图像监测传递系统与方法
CN115684688B (zh) * 2022-03-31 2024-02-06 江苏未来智慧信息科技有限公司 针对防窃电系统的图像监测传递系统与方法
CN114584617A (zh) * 2022-05-06 2022-06-03 南京维数软件股份有限公司 一种基于汇聚处理平台的抓拍数据分发方法及装置
CN114584617B (zh) * 2022-05-06 2022-08-02 南京维数软件股份有限公司 一种基于汇聚处理平台的抓拍数据分发方法及装置

Also Published As

Publication number Publication date
US20220279111A1 (en) 2022-09-01
EP4024275A1 (en) 2022-07-06
EP4024275A4 (en) 2022-11-09
CN112446395A (zh) 2021-03-05
US11750776B2 (en) 2023-09-05
CN112446395B (zh) 2023-07-25

Similar Documents

Publication Publication Date Title
WO2021036832A1 (zh) 网络摄像机、视频监控系统及方法
US9277165B2 (en) Video surveillance system and method using IP-based networks
EP2688296B1 (en) Video monitoring system and method
CN109299703B (zh) 对鼠情进行统计的方法、装置以及图像采集设备
WO2020094091A1 (zh) 一种图像抓拍方法、监控相机及监控系统
US7423669B2 (en) Monitoring system and setting method for the same
WO2020094088A1 (zh) 一种图像抓拍方法、监控相机及监控系统
CN106815575B (zh) 人脸检测结果集的优选系统及其方法
WO2020024749A1 (zh) 一种基于多机协作的图像处理方法和视频监控系统
CN113190700B (zh) 用于房产交易的人脸抓拍、筛选和存储的方法及系统
CN103974037A (zh) 一种基于网络视频云的实时检索系统及其方法
CN114863364B (zh) 一种基于智能视频监控的安防检测方法及系统
CN111050027A (zh) 镜头畸变补偿方法、装置、设备和存储介质
US20230260325A1 (en) Person category attribute-based remote care method and device, and readable storage medium
CN110569715A (zh) 一种基于卷积神经网络的人脸识别系统
CN113283410B (zh) 基于数据关联分析的人脸增强识别方法、装置和设备
CN115733957A (zh) 一种基于图像识别的导航设备告警处理方法
US10063421B2 (en) System and method for identifying imaging devices
CN114359828A (zh) 目标行为的记录方法、装置、存储介质及电子装置
CN112073676A (zh) 一种点名系统
JP6632632B2 (ja) 監視システム
CN116170494B (zh) 气象数据的可视化方法及系统
CN113591549B (zh) 一种视频事件检测方法、计算机设备以及装置
KR102637949B1 (ko) 썸네일을 관리하는 서버, 방법, 썸네일을 이용하는 단말
JP2018160820A (ja) 画像処理システム及び画像処理方法

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: 20857136

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2020857136

Country of ref document: EP

Effective date: 20220329