WO2020108573A1 - 视频图像遮挡方法、装置、设备及存储介质 - Google Patents

视频图像遮挡方法、装置、设备及存储介质 Download PDF

Info

Publication number
WO2020108573A1
WO2020108573A1 PCT/CN2019/121644 CN2019121644W WO2020108573A1 WO 2020108573 A1 WO2020108573 A1 WO 2020108573A1 CN 2019121644 W CN2019121644 W CN 2019121644W WO 2020108573 A1 WO2020108573 A1 WO 2020108573A1
Authority
WO
WIPO (PCT)
Prior art keywords
target
video image
targets
information
frame
Prior art date
Application number
PCT/CN2019/121644
Other languages
English (en)
French (fr)
Inventor
车军
陈畅怀
陆海先
任烨
朱江
Original Assignee
杭州海康威视数字技术股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 杭州海康威视数字技术股份有限公司 filed Critical 杭州海康威视数字技术股份有限公司
Publication of WO2020108573A1 publication Critical patent/WO2020108573A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T11/00—2D [Two Dimensional] image generation
    • G06T11/40—Filling a planar surface by adding surface attributes, e.g. colour or texture
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T3/00—Geometric image transformations in the plane of the image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/24—Aligning, centring, orientation detection or correction of the image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]

Definitions

  • the present disclosure relates to the technical field of video processing, and in particular, to a video image occlusion method, device, equipment, and storage medium.
  • a method for realizing video image occlusion in the related art is as follows: the video under the current monitoring scene is collected by a camera, and the same fixed area in each frame of the video image of the video is masked according to a pre-configured fixed area.
  • the above technology blocks the fixed area in the video image by configuring the fixed area, so as to achieve the effect of blocking the target in the fixed area.
  • the position of the target that needs to be blocked changes, such as moving out of the fixed area, when the fixed area in the video image is still blocked, the target that needs to be blocked will not be blocked, resulting in privacy leakage. Therefore, it is urgent A video image occlusion method is needed to accurately and effectively occlude the target that needs to be occluded.
  • Embodiments of the present disclosure provide a video image occlusion method, device, equipment, and storage medium, which can solve the problem that the target that needs to be occlusion in the related art is not occlusion.
  • the technical solution is as follows:
  • a video image occlusion method includes:
  • the motion trajectory information of each target appearing in the video based on the multi-frame video image of the video, the motion trajectory information of each target including the position information and size information of each target in the multi-frame video image;
  • the area where the first target is located in the multi-frame video image is blocked, and the area where the first target is located is position information and size information of the first target The corresponding area.
  • the multi-frame video image based on the video, and acquiring the movement track information of each target appearing in the video includes:
  • target detection is performed on the video image to determine multiple targets in the video image
  • the method further includes:
  • the method further includes:
  • the position information and size information of each object in the plurality of objects in the video image are stored in correspondence with the unique identifier of each object.
  • the method further includes:
  • the video image is a video image other than the first frame video image in the multi-frame video image
  • determine a known target and an unknown target among the multiple targets the known target is the video image
  • the method further includes:
  • the position information and size information of the unknown target in the video image are stored in correspondence with the unique identification of the unknown target.
  • the method further includes:
  • the image features of the multiple targets are stored in correspondence with the unique identifiers of the multiple targets.
  • the extracting image features of the plurality of targets in the video image includes:
  • the evaluation information including at least one of posture, size, imaging condition, occlusion and shooting angle;
  • the determining the first target that needs to be blocked among the respective targets includes:
  • the target corresponding to the first selection event is determined to be the first target, and the first selection event is used to select a target to be blocked from each of the targets;
  • a target other than the target corresponding to the second selection event is determined as the first target, and the second selection event is used to select from the various targets that do not need to be blocked aims.
  • the blocking the area where the first target is located in the multi-frame video image according to the movement trajectory information of the first target includes:
  • the region where the target and the first target are located in the multi-frame video image is blocked.
  • the goal of determining that the first goal is the same real goal as the first goal includes:
  • the respective targets and the first target determine the targets that are the same real target among the respective targets and the first target.
  • the determining, according to the similarity between each target and the first target, the target that is the same real target among the various targets and the first target includes:
  • the target corresponding to the target confirmation event is determined to be the same real target as the first target, and the target confirmation event is used to select from the respective targets
  • the first goal is the goal of the same real goal.
  • the determining, according to the similarity between each target and the first target, the target that is the same real target among the various targets and the first target includes:
  • the target whose similarity to the first target is greater than a preset threshold is determined to be the same real target as the first target.
  • a video image blocking device in a second aspect, includes:
  • An obtaining module used to obtain the motion trajectory information of each target appearing in the video based on the multi-frame video image of the video, and the motion trajectory information of each target includes the position of each target in the multi-frame video image Information and size information;
  • a determining module configured to determine a first target that needs to be occluded among the various targets
  • the occlusion module is configured to occlude the area where the first target is located in the multi-frame video image according to the movement trajectory information of the first target, and the area where the first target is located is the area of the first target The area corresponding to the location information and size information.
  • the acquiring module is configured to perform target detection on the video image for each frame of video images in the multi-frame video image to determine multiple targets in the video image; The position information and size information of the multiple objects in the video image.
  • the device further includes:
  • a first generating module configured to generate a unique identifier of each target in the plurality of targets when the video image is the first frame of video images in the multi-frame video images;
  • the first storage module is configured to store position information and size information of each of the plurality of objects in the video image in correspondence with the unique identification of each object.
  • the device further includes:
  • the determining module is further configured to determine a known target and an unknown target among the multiple targets when the video image is a video image other than the first frame video image in the multi-frame video image.
  • the known target is a target included in the previous video image of the video image
  • the unknown target is a target that is not included in the previous video image
  • a second generation module used to generate a unique identifier of the unknown target
  • a second storage module for storing the position information and size information of the known target in the video image in correspondence with the unique identification of the known target; storing the position of the unknown target in the video image The information and size information are stored in correspondence with the unique identification of the unknown target.
  • the device further includes:
  • An extraction module for extracting image features of the multiple targets in the video image
  • the third storage module is configured to store the image features of the multiple targets in correspondence with the unique identifiers of the multiple targets.
  • the extraction module is used to obtain evaluation information of the plurality of targets in the video image, where the evaluation information includes at least one of posture, size, imaging condition, occlusion, and shooting angle Select the target whose evaluation information meets the preset evaluation conditions from the multiple targets; extract the image features of the selected target.
  • the determination module is used to display the various targets; when a first selection event is detected, the target corresponding to the first selection event is determined as the first target, so The first selection event is used to select a target to be blocked from the respective targets; when a second selection event is detected, a target other than the target corresponding to the second selection event is determined as the first target, The second selection event is used to select a target that does not require occlusion from the respective targets.
  • the occlusion module is used to determine the target that is the same real target as the first target among the various targets; based on the target and the first target in the multi-frame video The position information and size information in the image block the area where the target and the first target are located in the multi-frame video image.
  • the determination module is used to compare the image features of the first target and the image features of the respective targets to obtain the similarity between the respective targets and the first target; According to the similarity between the respective targets and the first target, determine the targets that are the same real target among the respective targets and the first target.
  • the determination module is configured to arrange and display the various targets according to the similarity between the respective targets and the first target, and the greater the similarity, the higher the alignment; when detected In the target confirmation event, the target corresponding to the target confirmation event is determined to be the same target as the first target.
  • the target confirmation event is used to select the first target from the respective targets Goals for the same real goal.
  • the determining module is configured to determine, according to the similarity between each target and the first target, a target whose similarity to the first target is greater than a preset threshold as the The first goal is the goal of the same real goal.
  • an electronic device including a processor and a memory; the memory is used to store at least one instruction; the processor is used to execute at least one instruction stored on the memory to implement the first The method steps described in any aspect of the aspect.
  • a computer-readable storage medium where at least one instruction is stored in the computer-readable storage medium, and the at least one instruction is executed by a processor to implement any one of the implementation manners of the first aspect Method steps.
  • a computer program product containing instructions, which when run on a computer, causes the computer to implement the method steps described in any one of the implementation manners of the first aspect above.
  • the movement trajectory information of each target appearing in the video Since the movement trajectory information of the target may include the position information and size information of the target in the multi-frame video image, after determining the first target to be blocked, the first target The motion trajectory information of the multi-frame video image blocks the area where the first target is located, and the area where the first target is located in each frame of the video image may be different, so as to achieve accurate and effective shielding of the target that needs to be blocked.
  • FIG. 1 is a flowchart of a video image occlusion method provided by an embodiment of the present disclosure
  • FIG. 2 is a flowchart of a video image occlusion method provided by an embodiment of the present disclosure
  • FIG. 3 is a schematic flowchart of generating an occlusion video according to an embodiment of the present disclosure
  • FIG. 4 is a schematic structural diagram of a video image blocking device according to an embodiment of the present disclosure.
  • FIG. 5 is a schematic structural diagram of a video image occlusion device provided by an embodiment of the present disclosure.
  • FIG. 6 is a schematic structural diagram of a video image blocking device according to an embodiment of the present disclosure.
  • FIG. 7 is a schematic structural diagram of a video image blocking device according to an embodiment of the present disclosure.
  • FIG. 8 is a schematic structural diagram of an electronic device 800 provided by an embodiment of the present disclosure.
  • the video image occlusion method provided by an embodiment of the present disclosure may be performed by an electronic device.
  • the electronic device may be equipped with a camera, or may be connected to the camera through a data cable to perform video monitoring through the camera to obtain video, namely The electronic device may have a video capture function.
  • the electronic device can also receive the video to be processed from other front-end devices.
  • the electronic device may be a smart camera device, a computer device, or the like, which is not limited in the embodiments of the present disclosure.
  • FIG. 1 is a flowchart of a video image occlusion method provided by an embodiment of the present disclosure. Referring to Figure 1, the method includes:
  • the motion trajectory information of the first target block the area where the first target is located in the multi-frame video image, and the area where the first target is located corresponds to the position information and size information of the first target region.
  • the movement trajectory information of each target appearing in the video is acquired. Since the movement trajectory information of the target may include the position information and size information of the target in the multi-frame video image, the first target that needs to be blocked is determined Afterwards, the area where the first target is located in the multi-frame video image can be blocked according to the movement trajectory information of the first target, and the area where the first target is located in each frame of the video image can be different, so that the target that needs to be blocked Perform accurate and effective occlusion.
  • acquiring the movement track information of each target appearing in the video includes:
  • target detection is performed on the video image to determine multiple targets in the video image
  • the method further includes:
  • the method further includes:
  • the position information and size information of each object in the plurality of objects in the video image are stored in correspondence with the unique identification of each object.
  • the method further includes:
  • the video image is a video image other than the first frame video image in the multi-frame video image
  • determine a known target and an unknown target among the multiple targets the known target is the previous frame video of the video image
  • the target contained in the image, the unknown target is a target not included in the previous video image
  • the method further includes:
  • the position information and size information of the unknown target in the video image are stored in correspondence with the unique identification of the unknown target.
  • the method further includes:
  • the image features of the multiple targets are stored in correspondence with the unique identifiers of the multiple targets.
  • extracting image features of the multiple targets in the video image includes:
  • the evaluation information including at least one of posture, size, imaging condition, occlusion condition, and shooting angle;
  • determining the first target that needs to be blocked among the various targets includes:
  • the target corresponding to the first selection event is determined as the first target, and the first selection event is used to select a target that needs to be occluded from each target;
  • a target other than the target corresponding to the second selection event is determined as the first target, and the second selection event is used to select a target that does not require occlusion from the respective targets.
  • blocking the area where the first target is located in the multi-frame video image according to the movement trajectory information of the first target includes:
  • the region where the target and the first target are located in the multi-frame video image is blocked.
  • the determination that each of the goals is the same as the first goal with the first goal includes:
  • the target that is the same as the first target among the various targets is determined.
  • determining the target that is the same real target as the first target in each target includes:
  • the various targets are arranged and displayed, the greater the similarity, the higher the ranking;
  • the target corresponding to the target confirmation event is determined to be the same target as the first target, and the target confirmation event is used to select from the various targets the same as the first target The goal of a real goal.
  • determining the target that is the same real target as the first target in each target includes:
  • the target whose similarity to the first target is greater than a preset threshold is determined to be the same real target as the first target.
  • FIG. 2 is a flowchart of a video image occlusion method provided by an embodiment of the present disclosure. Referring to Figure 2, the method includes:
  • the motion trajectory information of each target includes the position information and size information of each target in the multi-frame video image.
  • Each target may include multiple types, for example, the type of the target may be a person, a face or a human body. Of course, the type of the target may also be a thing. The embodiment of the present disclosure does not specifically limit the type of the target.
  • the multi-frame video image may include all the video images in the video, or the multi-frame video image may also include part of the video images in the video.
  • the video may be extracted from the video according to a preset acquisition strategy Acquiring the multi-frame video image, for example, the first frame video image to the n-th frame video image in the video can be acquired to obtain the multi-frame video image, where n can be set according to actual needs.
  • this step 201 and subsequent steps may be performed by an electronic device.
  • the electronic device Before performing this step 201, the electronic device needs to acquire the multi-frame video image of the video. Taking the electronic device having a video collection function as an example, the electronic device can perform video collection on a certain monitoring area to obtain the video, and obtain multi-frame video images of the video. Of course, the electronic device can also receive the video sent by the front-end device (such as a surveillance camera), and acquire multi-frame video images of the video.
  • the front-end device such as a surveillance camera
  • this step 201 may include: for each video image in the multi-frame video image, performing target detection on the video image to determine multiple targets in the video image; acquiring the multiple targets Position information and size information in the video image.
  • the electronic device performs target detection on the first frame video image to determine the first frame After multiple targets in the video image, a unique identification of each of the multiple targets can be generated, and the unique identification can be represented by an ID (Identification, identification). After acquiring the position information and size information of multiple targets in the first frame of the video image, the electronic device may uniquely identify the position information and size information of each target of the multiple targets in the video image and each target Corresponding storage.
  • the electronic device For the video image currently subject to target detection, when the video image is a video image other than the first frame video image in the multi-frame video image, the electronic device performs target detection on the video image to determine how many in the video image After each target, the known target and the unknown target among the multiple targets can be determined, and a unique identification of the unknown target can be generated. After acquiring the position information and size information of multiple targets in the video image, the electronic device may store the position information and size information of the known target in the video image corresponding to the unique identifier of the known target, and store the unknown The position information and size information of the target in the video image are stored corresponding to the unique identification of the unknown target.
  • the known target is the target contained in the video image of the previous frame of the video image, that is, the target whose position information and size information in the previous frame of the video image have been obtained
  • the unknown target is the target A target that is not included in a frame of video image, that is, a target that has not obtained its position information and size information in the previous frame of video image.
  • the electronic device For the first frame of the multi-frame video image, the electronic device will generate a unique identifier for all detected targets after performing target detection on the first frame of the video image. For each frame of video image after the first frame of video image, after detecting the target of the current video image, the electronic device will determine which of all the detected targets are the targets that have been detected in the previous frame of video image (i.e. Known targets), which are the targets that were not detected in the previous video image (ie unknown targets), for unknown targets, the electronic device will consider them as a new target, so for this unknown target, a new unique logo.
  • the above process is actually the process of target detection and target tracking.
  • the target enters the video screen at the first moment and leaves the video screen at the second moment
  • the target can be detected in the video image collected at the first moment, and is collected at the moment between the first moment and the second moment
  • the target can also be detected in the obtained video image, but the target will not be detected in the video image collected at the second moment.
  • the electronic device may extract the video image from the video image.
  • Image features of multiple targets and store the image features of the multiple targets in correspondence with the unique identifiers of the multiple targets.
  • the electronic device may use a feature extraction model to extract image features that can describe the target in the image, and the image features may be represented by a string of binary codes.
  • the feature extraction model can be obtained by training with a large number of samples using machine learning methods.
  • the electronic device may extract image features for each of the multiple targets, or extract image features only for some of the targets. For example, the electronic device can first evaluate the target in the video image, and then extract the image feature of the target that meets the preset evaluation condition.
  • the target that meets the preset evaluation condition can generally extract the image feature that accurately describes the target. It can reduce the resource consumption caused by meaningless image feature extraction.
  • the preset evaluation condition can be set by the user according to actual needs, or can be set by the electronic device by default, which is not limited in the embodiments of the present disclosure.
  • the electronic device may obtain evaluation information of multiple targets in the video image.
  • the evaluation information includes posture, size, imaging conditions, occlusion, and At least one of the shooting angles; from the plurality of targets, select targets whose evaluation information meets preset evaluation conditions; and extract image features of the selected targets.
  • the posture refers to the posture of the target in the image, such as sitting, standing, etc.
  • the size refers to the size of the target imaged in the image
  • the imaging conditions can include whether the target is imaged in the image with or without shadow
  • the occlusion can be Including different degrees of occlusion, such as no occlusion, partial occlusion, and severe occlusion
  • the shooting angle can include shooting height, shooting direction, and shooting distance.
  • the target whose evaluation information meets the preset evaluation conditions is selected from the multiple targets, including: according to the evaluation information of the multiple targets and the multiple targets Type, select the goal that the evaluation information meets the preset evaluation conditions.
  • two types of targets, human face and human body can correspond to different preset evaluation conditions.
  • Each target in this step 201 may be each target to which all IDs belong, and the targets with different IDs in each target may be the same real target, that is, the same real target may have multiple IDs.
  • the electronic device considers the target as multiple targets and generates multiple IDs through the target tracking algorithm.
  • target A if target A enters the monitoring area at time t0, the electronic device generates ID1 for target A and tracks it. If the target A leaves the monitoring area at time t1, the position information, size information, and image characteristics of the target A in the video image collected before time t1 will be stored in correspondence with the ID1 as a target.
  • target A enters the monitoring area again, the electronic device considers the target A to be a new target, generates a new ID2 for the target A, tracks it and re-extracts image features.
  • target A may include two IDs ID1 and ID2.
  • each video image in multiple video images includes multiple targets.
  • some video images in the multiple video images may only include
  • the realization principle of a goal at this time is the same as or similar to the realization of the above-mentioned video image including multiple goals, that is, the same treatment can be performed on the goal according to the method provided by the embodiments of the present disclosure, which will not be described here.
  • determining the first target that needs to be blocked among the various targets includes: displaying each target; when a first selection event is detected, determining the target corresponding to the first selection event as the first target A target, the first selection event is used to select a target to be occluded from each target; when a second selection event is detected, a target other than the second target is determined as the first target, the second selection event It is used to select the second target that does not require occlusion from the respective targets.
  • each target in step 201 it may be each target to which all IDs belong.
  • a partial image of each target may be displayed on the user interaction interface, such as a partial image containing the target captured from a certain frame of video image. If the electronic device has no display function, the partial image of each target may be sent to the user device, and the user device displays the partial image of each target on the user interaction interface.
  • the electronic device interacts with the user and displays all the objects in the video to the user through the user interaction interface.
  • the user can browse to select the target that needs to be blocked or does not need to be blocked.
  • the electronic device can Select and determine a first target that needs to be blocked or a second target that does not need to be blocked.
  • the first target is an irrelevant target that the user does not want to pay attention to or is not interested in
  • the second target is a target that the user wants to pay attention to or interested in.
  • the electronic device considers the target to be multiple targets, and some of the targets may be the same real target.
  • the electronic device can find the target that is the same real target as the first target from the respective targets through target comparison.
  • the target comparison may be to use a preset calculation method to obtain the similarity between the image features of the two targets.
  • the electronic device may determine the target that is the same real target as the first target through the similarity between each target and the first target. Specifically, the electronic device may compare the image features of the first target and the image features of the respective targets to obtain the similarity between the respective targets and the first target; according to the similarity between the respective targets and the first target , Determine that each of the targets is the same as the first target.
  • the electronic device can obtain one or more image features of the first target stored corresponding to the ID according to the ID of the first target, and then compare the image features of all targets in the video with the first target.
  • the image features of a target are compared in pairs to calculate the similarity.
  • the electronic device may use the Euclidean distance to calculate the similarity between two image features. The smaller the Euclidean distance, the greater the similarity.
  • the calculation of the similarity is not limited to the Euclidean distance. For example, it may be a cosine similarity.
  • the embodiment of the present disclosure does not specifically limit the calculation method of the similarity.
  • the electronic device determines, according to the similarity between each target and the first target, that the target is the same real target as the first target among the various targets, including but not limited to the following two possible implementation modes:
  • the first way is to arrange and display each target according to the similarity between the target and the first target.
  • the target is determined to be the same real target as the first target, and the target confirmation event is used to select a target that is the same real target as the first target from the various targets.
  • the electronic device determines the target that is the same as the real target according to the user's confirmation operation. For the case where there are multiple IDs for a real target described in step 201, by displaying the compared targets in order of similarity from high to low, the user confirms whether the top target is based on the comparison result The same real target, and make selections, such as selecting multiple targets (multiple IDs) that are the same real target as the first target.
  • a target whose similarity to the first target is greater than a preset threshold is determined to be the same real target as the first target.
  • the preset threshold may be set by the user according to actual needs, or may be set by the electronic device by default, which is not limited in the embodiments of the present disclosure.
  • the electronic device determines the target that is the same real target as the first target according to the similarity between each target and the first target, which can reduce user operations.
  • the electronic device may consider that the target and the first target are the same real target, and are both targets that need to be blocked.
  • the area where the target is located is the area corresponding to the position information and size information of the target.
  • the electronic device can obtain the stored first target and the motion trajectories of all targets that are the same real target as the first target Information for privacy masking operations.
  • Privacy occlusion refers to the use of some technical means to block sensitive targets in pictures or videos.
  • the electronic device can block all targets that are the same real target as the first target in each frame of the multi-frame image, thereby generating a blocked video.
  • Blocked video refers to blocking a specific target in the original video After getting the video.
  • the electronic device may block the target in each frame of the video image according to the position information and size information of the target in each frame of the video image.
  • the blocking method includes but is not limited to superimposing an opaque blocking block on the corresponding area.
  • the blocking block may be set to a target color or a mosaic form as long as it can play a blocking role.
  • step 203 and step 204 are a possible implementation manner of blocking the area where the first target is located in the multi-frame video image according to the movement trajectory information of the first target, where the first target is located
  • the area of is the area corresponding to the position information and size information of the first target. By covering all the targets that are the same real target as the first target, the comprehensiveness of the blocking can be ensured.
  • occlusion of the area where the first target is located in the multi-frame video image may also include other possible implementation manners, for example, directly based on the first The motion trajectory information of the target determines the area to be blocked in the multi-frame video image, and then blocks the determined area in the multi-frame video image.
  • FIG. 3 is a schematic flowchart of generating an occlusion video according to an embodiment of the present disclosure.
  • the entire process may be implemented by a data extraction unit, a data storage unit, an occlusion processing unit, and a user interaction unit, where the data extraction unit includes target detection, target tracking, target evaluation, and target feature extraction of video.
  • the data extraction unit is responsible for extracting all the targets in a given video and tracking them.
  • Each tracked target generates a unique identification ID.
  • each target is tracked in time sequence. For example, each target is acquired in each The position and size in the frame video image. During target tracking, the target is evaluated in real time.
  • the evaluation indicators for real-time evaluation may include but not limited to posture, size, imaging conditions, occlusion factors, angle, etc., and select one that meets the evaluation conditions according to the target type Or multiple targets to extract image features.
  • the data storage unit may store data in the form of metadata (Metadata), and store the ID, motion track information, and image characteristics of each target extracted by the data extraction unit.
  • the occlusion processing unit combined with the user interaction unit provides a semi-automatic privacy occlusion operation, including providing a target preview, the user selects the target of interest, obtains the image features of the target, performs feature comparison, manually compares and screens the results, and then obtains The trajectory information of the target finally generates an occlusion video.
  • the embodiment of the present disclosure takes steps 201 to 204 as an example for execution by an electronic device, that is, the electronic device may be integrated with the data extraction unit, data storage unit, and occlusion processing unit in FIG. 3 Various functions realized by the user interaction unit.
  • the steps 201 to 204 may also be performed by different devices, that is, the functions implemented by the units in FIG. 3 may be integrated on different devices, respectively.
  • a simple occlusion method that allows the same real target to enter and exit the same scene multiple times can also be achieved.
  • the method can also be used to occlusion the same target in different scenarios. For example, if the different scenes may be scenes where different cameras shoot different videos in the same monitoring area, different videos for different scenes can be blocked by the above steps 201 to 204 to the same target in the monitoring area.
  • the technical solution provided by the embodiments of the present disclosure can significantly increase the amount of video processing Reduce user manual operation costs.
  • the movement trajectory information of each target appearing in the video is acquired. Since the movement trajectory information of the target may include the position information and size information of the target in the multi-frame video image, the first target that needs to be blocked is determined Afterwards, the area where the first target is located in the multi-frame video image can be blocked according to the movement trajectory information of the first target, and the area where the first target is located in each frame of the video image can be different, so that the target that needs to be blocked Perform accurate and effective occlusion.
  • FIG. 4 is a schematic structural diagram of a video image blocking device provided by an embodiment of the present disclosure. 4, the device includes:
  • the obtaining module 401 is used to obtain the motion trajectory information of each target appearing in the video based on the multi-frame video image of the video, and the motion trajectory information of each target includes the position information and size of each target in the multi-frame video image information;
  • the determining module 402 is used to determine the first target that needs to be occluded among the various targets;
  • the occlusion module 403 is configured to occlude the area where the first target is located in the multi-frame video image according to the movement track information of the first target, and the area where the first target is located is the position information and size of the first target The area corresponding to the information.
  • the acquisition module is used to perform target detection on each video image in the multi-frame video image to determine multiple targets in the video image; acquire the multiple targets in Position information and size information in the video image.
  • the device further includes:
  • the first generating module 404 is configured to generate a unique identifier of each target in the multiple targets when the video image is the first frame of the multiple frame video images;
  • the first storage module 405 is configured to store position information and size information of each of the plurality of objects in the video image in correspondence with the unique identification of each object.
  • the device further includes:
  • the determining module 402 is further configured to determine a known target and an unknown target among the multiple targets when the video image is a video image other than the first frame video image in the multi-frame video image, the known target is the The target contained in the previous video image of the video image, the unknown target is a target not included in the previous video image;
  • the second generation module 406 is used to generate a unique identifier of the unknown target
  • the second storage module 407 is configured to store the position information and size information of the known target in the video image corresponding to the unique identification of the known target; the position information and size information of the unknown target in the video image Store corresponding to the unique identifier of the unknown target.
  • the device further includes:
  • An extraction module 408, configured to extract image features of the multiple targets in the video image
  • the third storage module 409 is configured to store the image features of the multiple targets in correspondence with the unique identifiers of the multiple targets.
  • the extraction module 408 is used to obtain evaluation information of the plurality of targets in the video image, the evaluation information including at least one of posture, size, imaging condition, occlusion, and shooting angle; from the Among multiple targets, select targets whose evaluation information meets preset evaluation conditions; extract image features of the selected targets.
  • the determination module 402 is used to display each target; when a first selection event is detected, the target corresponding to the first selection event is determined as the first target, the first selection The event is used to select a target to be blocked from the various targets; when a second selection event is detected, a target other than the target corresponding to the second selection event is determined as the first target, and the second selection event is used to Select the target that does not require occlusion from each target.
  • the occlusion module 403 is used to determine the target that is the same real target as the first target in each target; according to the position information of the target and the first target in the multi-frame video image And size information, to block the area where the target and the first target are located in the multi-frame video image.
  • the determination module 402 is used to compare the image features of the first target and the image features of the respective targets to obtain the similarity between the respective targets and the first target; according to the respective targets The similarity to the first target determines the target that is the same real target among the various targets and the first target.
  • the determination module 402 is used to arrange and display the various targets according to the similarity between the respective targets and the first target, and the greater the similarity, the higher the alignment; when a target confirmation event is detected At this time, the target corresponding to the target confirmation event is determined to be the same real target as the first target, and the target confirmation event is used to select a target that is the same real target as the first target from the various targets.
  • the determination module 402 is configured to determine, according to the similarity of the various targets and the first target, a target whose similarity to the first target is greater than a preset threshold as the first target The goal of the same real goal.
  • the movement trajectory information of each target appearing in the video is acquired. Since the movement trajectory information of the target may include the position information and size information of the target in the multi-frame video image, the first target that needs to be blocked is determined Afterwards, the area where the first target is located in the multi-frame video image can be blocked according to the movement trajectory information of the first target, and the area where the first target is located in each frame of the video image can be different, so that the target that needs to be blocked Perform accurate and effective occlusion.
  • the video image occlusion device provided in the above embodiments only uses the division of the above functional modules as an example to illustrate the video image occlusion.
  • the above functions can be allocated by different functional modules according to needs. That is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above.
  • the video image occlusion device and the video image occlusion method embodiment provided in the above embodiments belong to the same concept. For the specific implementation process, see the method embodiments, and details are not described here.
  • the electronic device 800 may have a relatively large difference due to different configurations or performances, and may include one or more processors (Central Processing Units, CPU). 801 and one or more memories 802, where at least one instruction is stored in the memory 802, and the at least one instruction is loaded and executed by the processor 801 to implement the video image occlusion method provided by the foregoing method embodiments.
  • the electronic device 800 may also have components such as a wired or wireless network interface, a keyboard, and an input-output interface for input and output.
  • the electronic device 800 may also include other components for implementing device functions, which will not be repeated here.
  • a computer-readable storage medium storing at least one instruction, for example, a memory storing at least one instruction, and when the at least one instruction is executed by a processor, the video image in the above embodiment is implemented Occlusion method.
  • the computer-readable storage medium may be read-only memory (Read-Only Memory, ROM), random-access memory (Random Access Memory, RAM), read-only compact disc (Compact Disc Read-Only Memory, CD-ROM), Magnetic tapes, floppy disks, optical data storage devices, etc.
  • the above program may be stored in a computer-readable storage medium.
  • the storage medium can be read-only memory, magnetic disk or optical disk.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Multimedia (AREA)
  • Image Analysis (AREA)

Abstract

一种视频图像遮挡方法、装置、设备及存储介质，属于视频处理技术领域。所述方法包括：基于视频的多帧视频图像，获取该视频内出现的各个目标的运动轨迹信息，该运动轨迹信息包括该各个目标在该多帧视频图像中的位置信息和尺寸信息（101）；确定该各个目标中需要遮挡的第一目标（102）；根据该第一目标的运动轨迹信息，对该多帧视频图像中该第一目标所在的区域进行遮挡，该第一目标所在的区域为该第一目标的位置信息和尺寸信息所对应的区域（103）。所述方法可以实现对需要遮挡的目标进行准确有效的遮挡。

Description

视频图像遮挡方法、装置、设备及存储介质
本公开要求于2018年11月28日提交的申请号为201811435929.1、发明名称为“视频图像遮挡方法及装置”的中国专利申请的优先权，其全部内容通过引用结合在本公开中。
技术领域
本公开涉及视频处理技术领域，尤其涉及一种视频图像遮挡方法、装置、设备及存储介质。
背景技术
随着近年来安防产业的发展，视频监控已经遍及我们生活的方方面面，我们在享受视频监控带来的安全的同时，对于隐私保护的需求也越来越迫切，尤其当一些视频在网络上或电视上公开时，需要对视频图像中的某些目标进行遮挡，防止隐私泄露的事件发生。
目前，相关技术实现视频图像遮挡的方法如下：通过摄像机采集当前监控场景下的视频，根据预先配置的固定区域，对该视频的每帧视频图像中的同一固定区域进行遮挡。
上述技术通过配置固定区域的方式，对视频图像中的固定区域进行遮挡，从而达到对固定区域中目标的遮挡效果。但是，如果需要遮挡的目标发生位置变化，如移出该固定区域，此时仍然对视频图像中的固定区域进行遮挡时，会导致需要遮挡的目标未被遮挡，从而发生隐私泄露问题，因此，亟需一种视频图像遮挡方法，来对需要遮挡的目标进行准确有效的遮挡。
发明内容
本公开实施例提供了一种视频图像遮挡方法、装置、设备及存储介质，可以解决相关技术需要遮挡的目标未被遮挡的问题。所述技术方案如下：
第一方面，提供了一种视频图像遮挡方法，所述方法包括：
基于视频的多帧视频图像，获取所述视频内出现的各个目标的运动轨迹信息，所述各个目标的运动轨迹信息包括所述各个目标在所述多帧视频图像中的 位置信息和尺寸信息；
确定所述各个目标中需要遮挡的第一目标；
根据所述第一目标的运动轨迹信息，对所述多帧视频图像中所述第一目标所在的区域进行遮挡，所述第一目标所在的区域为所述第一目标的位置信息和尺寸信息所对应的区域。
在一种可能实现方式中，所述基于视频的多帧视频图像，获取所述视频内出现的各个目标的运动轨迹信息，包括：
对于所述多帧视频图像中的每帧视频图像，对所述视频图像进行目标检测，确定所述视频图像中的多个目标；
获取所述多个目标在所述视频图像中的位置信息和尺寸信息。
在一种可能实现方式中，所述对所述视频图像进行目标检测，确定所述视频图像中的多个目标之后，所述方法还包括：
当所述视频图像为所述多帧视频图像中的第一帧视频图像时，生成所述多个目标中每个目标的唯一标识；
所述获取所述多个目标在所述视频图像中的位置信息和尺寸信息之后，所述方法还包括：
将所述多个目标中每个目标在所述视频图像中的位置信息和尺寸信息与所述每个目标的唯一标识对应存储。
在一种可能实现方式中，所述对所述视频图像进行目标检测，确定所述视频图像中的多个目标之后，所述方法还包括：
当所述视频图像为所述多帧视频图像中除第一帧视频图像以外的视频图像时，确定所述多个目标中的已知目标和未知目标，所述已知目标为所述视频图像的上一帧视频图像中包含的目标，所述未知目标为所述上一帧视频图像中未包含的目标；
生成所述未知目标的唯一标识；
所述获取所述多个目标在所述视频图像中的位置信息和尺寸信息之后，所述方法还包括：
将所述已知目标在所述视频图像中的位置信息和尺寸信息与所述已知目标的唯一标识对应存储；
将所述未知目标在所述视频图像中的位置信息和尺寸信息与所述未知目标的唯一标识对应存储。
在一种可能实现方式中，所述对所述视频图像进行目标检测，确定所述视频图像中的多个目标之后，所述方法还包括：
提取所述视频图像中所述多个目标的图像特征；
将所述多个目标的图像特征与所述多个目标的唯一标识对应存储。
在一种可能实现方式中，所述提取所述视频图像中所述多个目标的图像特征，包括：
获取所述视频图像中所述多个目标的评价信息，所述评价信息包括姿态、尺寸、成像条件、遮挡情况和拍摄角度中至少一项；
从所述多个目标中，选取评价信息满足预设评价条件的目标；
提取所选取的目标的图像特征。
在一种可能实现方式中，所述确定所述各个目标中需要遮挡的第一目标，包括：
将所述各个目标进行展示；
当检测到第一选择事件时，将所述第一选择事件对应的目标确定为所述第一目标，所述第一选择事件用于从所述各个目标中选择需要遮挡的目标；
当检测到第二选择事件时，将除所述第二选择事件对应的目标以外的目标确定为所述第一目标，所述第二选择事件用于从所述各个目标中选择不需要遮挡的目标。
在一种可能实现方式中，所述根据所述第一目标的运动轨迹信息，对所述多帧视频图像中所述第一目标所在的区域进行遮挡，包括：
确定所述各个目标中与所述第一目标为同一个真实目标的目标；
根据所述目标和所述第一目标在所述多帧视频图像中的位置信息和尺寸信息，对所述多帧视频图像中所述目标和所述第一目标所在的区域进行遮挡。
在一种可能实现方式中，所述确定所述各个目标中与所述第一目标为同一个真实目标的目标，包括：
将所述第一目标的图像特征和所述各个目标的图像特征进行比对，获取所述各个目标与所述第一目标的相似度；
根据所述各个目标与所述第一目标的相似度，确定所述各个目标中与所述第一目标为同一个真实目标的目标。
在一种可能实现方式中，所述根据所述各个目标与所述第一目标的相似度，确定所述各个目标中与所述第一目标为同一个真实目标的目标，包括：
根据所述各个目标与所述第一目标的相似度，将所述各个目标进行排列展示，相似度越大排列越靠前；
当检测到目标确认事件时，将所述目标确认事件对应的目标确定为与所述第一目标为同一个真实目标的目标，所述目标确认事件用于从所述各个目标中选择与所述第一目标为同一个真实目标的目标。
在一种可能实现方式中，所述根据所述各个目标与所述第一目标的相似度，确定所述各个目标中与所述第一目标为同一个真实目标的目标，包括：
根据所述各个目标与所述第一目标的相似度，将与所述第一目标的相似度大于预设阈值的目标确定为与所述第一目标为同一个真实目标的目标。
第二方面，提供了一种视频图像遮挡装置，所述装置包括：
获取模块，用于基于视频的多帧视频图像，获取所述视频内出现的各个目标的运动轨迹信息，所述各个目标的运动轨迹信息包括所述各个目标在所述多帧视频图像中的位置信息和尺寸信息；
确定模块，用于确定所述各个目标中需要遮挡的第一目标；
遮挡模块，用于根据所述第一目标的运动轨迹信息，对所述多帧视频图像中所述第一目标所在的区域进行遮挡，所述第一目标所在的区域为所述第一目标的位置信息和尺寸信息所对应的区域。
在一种可能实现方式中，所述获取模块用于对于所述多帧视频图像中的每帧视频图像，对所述视频图像进行目标检测，确定所述视频图像中的多个目标；获取所述多个目标在所述视频图像中的位置信息和尺寸信息。
在一种可能实现方式中，所述装置还包括：
第一生成模块，用于当所述视频图像为所述多帧视频图像中的第一帧视频图像时，生成所述多个目标中每个目标的唯一标识；
第一存储模块，用于将所述多个目标中每个目标在所述视频图像中的位置信息和尺寸信息与所述每个目标的唯一标识对应存储。
在一种可能实现方式中，所述装置还包括：
所述确定模块还用于当所述视频图像为所述多帧视频图像中除第一帧视频图像以外的视频图像时，确定所述多个目标中的已知目标和未知目标，所述已知目标为所述视频图像的上一帧视频图像中包含的目标，所述未知目标为所述上一帧视频图像中未包含的目标；
第二生成模块，用于生成所述未知目标的唯一标识；
第二存储模块，用于将所述已知目标在所述视频图像中的位置信息和尺寸信息与所述已知目标的唯一标识对应存储；将所述未知目标在所述视频图像中的位置信息和尺寸信息与所述未知目标的唯一标识对应存储。
在一种可能实现方式中，所述装置还包括：
提取模块，用于提取所述视频图像中所述多个目标的图像特征；
第三存储模块，用于将所述多个目标的图像特征与所述多个目标的唯一标识对应存储。
在一种可能实现方式中，所述提取模块用于获取所述视频图像中所述多个目标的评价信息，所述评价信息包括姿态、尺寸、成像条件、遮挡情况和拍摄角度中至少一项；从所述多个目标中，选取评价信息满足预设评价条件的目标；提取所选取的目标的图像特征。
在一种可能实现方式中，所述确定模块用于将所述各个目标进行展示；当检测到第一选择事件时，将所述第一选择事件对应的目标确定为所述第一目标，所述第一选择事件用于从所述各个目标中选择需要遮挡的目标；当检测到第二选择事件时，将除所述第二选择事件对应的目标以外的目标确定为所述第一目标，所述第二选择事件用于从所述各个目标中选择不需要遮挡的目标。
在一种可能实现方式中，所述遮挡模块用于确定所述各个目标中与所述第一目标为同一个真实目标的目标；根据所述目标和所述第一目标在所述多帧视频图像中的位置信息和尺寸信息，对所述多帧视频图像中所述目标和所述第一目标所在的区域进行遮挡。
在一种可能实现方式中，所述确定模块用于将所述第一目标的图像特征和所述各个目标的图像特征进行比对，获取所述各个目标与所述第一目标的相似度；根据所述各个目标与所述第一目标的相似度，确定所述各个目标中与所述第一目标为同一个真实目标的目标。
在一种可能实现方式中，所述确定模块用于根据所述各个目标与所述第一目标的相似度，将所述各个目标进行排列展示，相似度越大排列越靠前；当检测到目标确认事件时，将所述目标确认事件对应的目标确定为与所述第一目标为同一个真实目标的目标，所述目标确认事件用于从所述各个目标中选择与所述第一目标为同一个真实目标的目标。
在一种可能实现方式中，所述确定模块用于根据所述各个目标与所述第一目标的相似度，将与所述第一目标的相似度大于预设阈值的目标确定为与所述 第一目标为同一个真实目标的目标。
第三方面，提供了一种电子设备，包括处理器和存储器；所述存储器，用于存放至少一条指令；所述处理器，用于执行所述存储器上所存放的至少一条指令，实现第一方面任一种实现方式所述的方法步骤。
第四方面，提供了一种计算机可读存储介质，所述计算机可读存储介质内存储有至少一条指令，所述至少一条指令被处理器执行时实现第一方面任一种实现方式所述的方法步骤。
第五方面，提供了一种包含指令的计算机程序产品，当其在计算机上运行时，使得计算机实现上述第一方面任一种实现方式所述的方法步骤。
本公开实施例提供的技术方案带来的有益效果至少包括：
获取视频内出现的各个目标的运动轨迹信息，由于目标的运动轨迹信息可以包括目标在多帧视频图像中的位置信息和尺寸信息，因而在确定需要遮挡的第一目标后，可以根据第一目标的运动轨迹信息，对该多帧视频图像中该第一目标所在区域进行遮挡，每帧视频图像中该第一目标所在的区域可以不同，从而实现对需要遮挡的目标进行准确有效的遮挡。
附图说明
为了更清楚地说明本公开实施例中的技术方案，下面将对实施例描述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公开的一些实施例，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是本公开实施例提供的一种视频图像遮挡方法的流程图；
图2是本公开实施例提供的一种视频图像遮挡方法的流程图；
图3是本公开实施例提供的一种生成遮挡视频的流程示意图；
图4是本公开实施例提供的一种视频图像遮挡装置的结构示意图；
图5是本公开实施例提供的一种视频图像遮挡装置的结构示意图；
图6是本公开实施例提供的一种视频图像遮挡装置的结构示意图；
图7是本公开实施例提供的一种视频图像遮挡装置的结构示意图；
图8是本公开实施例提供的一种电子设备800的结构示意图。
具体实施方式
为使本公开的目的、技术方案和优点更加清楚，下面将结合附图对本公开实施方式作进一步地详细描述。
在对本公开实施例提供的视频图像遮挡方法进行详细介绍之前，先对本公开实施例涉及的执行主体进行简单介绍。本公开实施例提供的视频图像遮挡方法可以由电子设备来执行，作为一种示例，该电子设备可以配置有摄像头，或者可以通过数据线与摄像头连接，以通过摄像头进行视频监控，得到视频，即该电子设备可以具有视频采集功能。或者，该电子设备还可以从其他前端设备接收待处理的视频。在一些实施例中，该电子设备可以为智能摄像设备、计算机设备等，本公开实施例对此不做限定。
在介绍完本公开实施例涉及的执行主体后，接下来将结合附图对本公开实施例提供的视频图像遮挡方法进行详细介绍。
图1是本公开实施例提供的一种视频图像遮挡方法的流程图。参见图1，该方法包括：
101、基于视频的多帧视频图像，获取该视频内出现的各个目标的运动轨迹信息，该各个目标的运动轨迹信息包括该各个目标在该多帧视频图像中的位置信息和尺寸信息。
102、确定该各个目标中需要遮挡的第一目标。
103、根据该第一目标的运动轨迹信息，对该多帧视频图像中该第一目标所在的区域进行遮挡，该第一目标所在的区域为该第一目标的位置信息和尺寸信息所对应的区域。
在本公开实施例中，获取视频内出现的各个目标的运动轨迹信息，由于目标的运动轨迹信息可以包括目标在多帧视频图像中的位置信息和尺寸信息，因而在确定需要遮挡的第一目标后，可以根据第一目标的运动轨迹信息，对该多帧视频图像中该第一目标所在区域进行遮挡，每帧视频图像中该第一目标所在的区域可以不同，从而实现对需要遮挡的目标进行准确有效的遮挡。
在一种可能实现方式中，基于视频的多帧视频图像，获取该视频内出现的各个目标的运动轨迹信息，包括：
对于该多帧视频图像中的每帧视频图像，对该视频图像进行目标检测，确定该视频图像中的多个目标；
获取该多个目标在该视频图像中的位置信息和尺寸信息。
在一种可能实现方式中，对该视频图像进行目标检测，确定该视频图像中的多个目标之后，该方法还包括：
当该视频图像为该多帧视频图像中的第一帧视频图像时，生成该多个目标中每个目标的唯一标识；
该获取该多个目标在该视频图像中的位置信息和尺寸信息之后，该方法还包括：
将该多个目标中每个目标在该视频图像中的位置信息和尺寸信息与该每个目标的唯一标识对应存储。
在一种可能实现方式中，对该视频图像进行目标检测，确定该视频图像中的多个目标之后，该方法还包括：
当该视频图像为该多帧视频图像中除第一帧视频图像以外的视频图像时，确定该多个目标中的已知目标和未知目标，该已知目标为该视频图像的上一帧视频图像中包含的目标，该未知目标为该上一帧视频图像中未包含的目标；
生成该未知目标的唯一标识；
该获取该多个目标在该视频图像中的位置信息和尺寸信息之后，该方法还包括：
将该已知目标在该视频图像中的位置信息和尺寸信息与该已知目标的唯一标识对应存储；
将该未知目标在该视频图像中的位置信息和尺寸信息与该未知目标的唯一标识对应存储。
在一种可能实现方式中，对该视频图像进行目标检测，确定该视频图像中的多个目标之后，该方法还包括：
提取该视频图像中该多个目标的图像特征；
将该多个目标的图像特征与该多个目标的唯一标识对应存储。
在一种可能实现方式中，提取该视频图像中该多个目标的图像特征，包括：
获取该视频图像中该多个目标的评价信息，该评价信息包括姿态、尺寸、成像条件、遮挡情况和拍摄角度中至少一项；
从该多个目标中，选取评价信息满足预设评价条件的目标；
提取所选取的目标的图像特征。
在一种可能实现方式中，确定该各个目标中需要遮挡的第一目标，包括：
将该各个目标进行展示；
当检测到第一选择事件时，将该第一选择事件对应的目标确定为该第一目标，该第一选择事件用于从该各个目标中选择需要遮挡的目标；
当检测到第二选择事件时，将除该第二选择事件对应的目标以外的目标确定为该第一目标，该第二选择事件用于从该各个目标中选择不需要遮挡的目标。
在一种可能实现方式中，根据该第一目标的运动轨迹信息，对该多帧视频图像中该第一目标所在的区域进行遮挡，包括：
确定该各个目标中与该第一目标为同一个真实目标的目标；
根据该目标和该第一目标在该多帧视频图像中的位置信息和尺寸信息，对该多帧视频图像中该目标和该第一目标所在的区域进行遮挡。
在一种可能实现方式中，确定该各个目标中与该第一目标为同一个真实目标的目标，包括：
将该第一目标的图像特征和该各个目标的图像特征进行比对，获取该各个目标与该第一目标的相似度；
根据该各个目标与该第一目标的相似度，确定该各个目标中与该第一目标为同一个真实目标的目标。
在一种可能实现方式中，根据该各个目标与该第一目标的相似度，确定该各个目标中与该第一目标为同一个真实目标的目标，包括：
根据该各个目标与该第一目标的相似度，将该各个目标进行排列展示，相似度越大排列越靠前；
当检测到目标确认事件时，将该目标确认事件对应的目标确定为与该第一目标为同一个真实目标的目标，该目标确认事件用于从该各个目标中选择与该第一目标为同一个真实目标的目标。
在一种可能实现方式中，根据该各个目标与该第一目标的相似度，确定该各个目标中与该第一目标为同一个真实目标的目标，包括：
根据该各个目标与该第一目标的相似度，将与该第一目标的相似度大于预设阈值的目标确定为与该第一目标为同一个真实目标的目标。
上述所有可选技术方案，可以采用任意结合形成本公开的可选实施例，在此不再一一赘述。
图2是本公开实施例提供的一种视频图像遮挡方法的流程图。参见图2，该 方法包括：
201、基于视频的多帧视频图像，获取该视频内出现的各个目标的运动轨迹信息，各个目标的运动轨迹信息包括该各个目标在该多帧视频图像中的位置信息和尺寸信息。
其中，各个目标可以包括多种类型，如目标的类型可以为人，也可以为人脸和人体，当然，目标的类型也可以为物，本公开实施例对目标的类型不做具体限定。
其中，该多帧视频图像可以包括该视频中所有的视频图像，或者，该多帧视频图像也可以包括该视频中的部分视频图像，在实施中，可以根据预先设定的获取策略从视频中获取该多帧视频图像，示例性地，可以获取该视频中的第一帧视频图像至第n帧视频图像，得到该多帧视频图像，其中，该n可以根据实际需求进行设置。
在一种可能实现方式中，该步骤201以及后续步骤可以由电子设备执行。
在执行该步骤201之前，电子设备需要先获取到该视频的多帧视频图像。以该电子设备具有视频采集功能为例，该电子设备可以对某个监控区域进行视频采集，得到该视频，并获取该视频的多帧视频图像。当然，电子设备也可以接收前端设备(如监控摄像头)发送的视频，并获取该视频的多帧视频图像。
在一种可能实现方式中，该步骤201可以包括：对于该多帧视频图像中的每帧视频图像，对该视频图像进行目标检测，确定该视频图像中的多个目标；获取该多个目标在该视频图像中的位置信息和尺寸信息。
具体地，对于当前进行目标检测的视频图像，当该视频图像为该多帧视频图像中的第一帧视频图像时，电子设备在对该第一帧视频图像进行目标检测，确定该第一帧视频图像中的多个目标之后，可以生成该多个目标中每个目标的唯一标识，该唯一标识可以用ID(Identification，身份标识)来表示。电子设备在获取该第一帧视频图像中多个目标的位置信息和尺寸信息后，可以将该多个目标中每个目标在该视频图像中的位置信息和尺寸信息与每个目标的唯一标识对应存储。
对于当前进行目标检测的视频图像，当该视频图像为该多帧视频图像中除第一帧视频图像以外的视频图像时，电子设备在对该视频图像进行目标检测，确定该视频图像中的多个目标之后，可以确定该多个目标中的已知目标和未知目标，并生成该未知目标的唯一标识。电子设备在获取该视频图像中多个目标 的位置信息和尺寸信息后，可以将该已知目标在该视频图像中的位置信息和尺寸信息与该已知目标的唯一标识对应存储，将该未知目标在该视频图像中的位置信息和尺寸信息与该未知目标的唯一标识对应存储。其中，该已知目标为该视频图像的上一帧视频图像中包含的目标，也即是，已获取其在上一帧视频图像中的位置信息和尺寸信息的目标，该未知目标为该上一帧视频图像中未包含的目标，也即是，未获取其在上一帧视频图像中的位置信息和尺寸信息的目标。
对于该多帧视频图像中的第一帧视频图像，电子设备在对该第一帧视频图像进行目标检测后，会为检测到的所有目标各自生成一个唯一标识。而对于第一帧视频图像以后的各帧视频图像，电子设备在对当前视频图像进行目标检测后，会判断检测出的所有目标中哪些是已在上一帧视频图像中检测到的目标(即已知目标)，哪些是未在上一帧视频图像中检测到的目标(即未知目标)，对于未知目标，电子设备会认为是一个新的目标，因而针对该未知目标，生成一个新的唯一标识。
上述过程实际上是进行目标检测和目标跟踪的过程。当目标在第一时刻进入视频画面，在第二时刻离开视频画面，则在第一时刻采集到的视频图像中可以检测到该目标，并且，在第一时刻和第二时刻之间的时刻采集到的视频图像中也可以检测到该目标，而在第二时刻采集到的视频图像中则不会检测到该目标。
在一种可能实现方式中，对于该多帧视频图像中的每帧视频图像，电子设备在对该视频图像进行目标检测，确定该视频图像中的多个目标之后，可以提取该视频图像中该多个目标的图像特征，并将该多个目标的图像特征与该多个目标的唯一标识对应存储。例如，电子设备可以采用特征提取模型，来提取图像中能够描述目标的图像特征，该图像特征可以用一串2进制码来表示。该特征提取模型可以采用机器学习的方法，通过大量样本训练得到。
具体地，电子设备可以对该多个目标中的每个目标均进行图像特征的提取，也可以仅对其中的某一些目标进行图像特征的提取。例如，电子设备可以先对视频图像中的目标进行评价后，再对满足预设评价条件的目标进行图像特征提取，满足预设评价条件的目标一般能够提取到准确完整描述该目标的图像特征，可以减少无意义的图像特征提取所带来的资源消耗。
其中，该预设评价条件由用户可以根据实际需求进行设置，或者，也可以由该电子设备默认设置，本公开实施例对此不做限定。
在一种可能实现方式中，对于该多帧视频图像中的每帧视频图像，电子设备可以获取该视频图像中多个目标的评价信息，该评价信息包括姿态、尺寸、成像条件、遮挡情况和拍摄角度中至少一项；从该多个目标中，选取评价信息满足预设评价条件的目标；提取所选取的目标的图像特征。其中，姿态是指图像中目标所处的姿态，如坐着、站着等；尺寸是指目标在图像中成像的尺寸；成像条件可以包括目标在图像中的成像有无阴影等；遮挡情况可以包括不同程度的遮挡，如无遮挡、部分遮挡和严重遮挡等；拍摄角度可以包括拍摄高度、拍摄方向和拍摄距离等。
不同类型的目标可以对应不同的预设评价条件，相应地，该从该多个目标中，选取评价信息满足预设评价条件的目标，包括：根据该多个目标的评价信息和该多个目标的类型，选取评价信息满足预设评价条件的目标。例如，人脸和人体这两种类型的目标可以对应不同的预设评价条件。
该步骤201中的各个目标可以是所有ID所属的各个目标，该各个目标中不同ID的目标可能为同一个真实目标，也即是，同一个真实目标可能具有多个ID。以该视频为监控区域的视频为例，当同一目标多次进出监控区域时，电子设备会认为该目标为多个目标，通过目标跟踪算法，生成多个ID。以目标A为例，如果目标A在t0时刻进入该监控区域，电子设备会针对该目标A生成ID1，并对其进行跟踪。如果在t1时刻，目标A离开该监控区域，则t1时刻之前采集到的视频图像中该目标A的位置信息、尺寸信息以及图像特征与该ID1会作为一个目标对应存储。如果在t2时刻，目标A再次进入该监控区域，则电子设备会认为该目标A是一个新的目标，针对该目标A生成一个新的ID2，并对其进行跟踪以及重新提取图像特征。这样，目标A可以包括ID1和ID2这两个ID。
需要说明的是，上述仅是以多个视频图像中的每个视频图像包括多个目标为例进行说明，在另一实施例中，该多个视频图像中的某些视频图像还可能仅包括一个目标，此时的实现原理与上述视频图像包括多个目标的实现相同或类似，即同样可以按照本公开实施例提供的方式对该一个目标进行相应的处理，这里不做过多介绍。
202、确定该各个目标中需要遮挡的第一目标。
在一种可能实现方式中，确定该各个目标中需要遮挡的第一目标包括：将该各个目标进行展示；当检测到第一选择事件时，将该第一选择事件对应的目标确定为该第一目标，该第一选择事件用于从该各个目标中选择需要遮挡的目 标；当检测到第二选择事件时，将除第二目标以外的目标确定为该第一目标，该第二选择事件用于从该各个目标中选择不需要遮挡的该第二目标。
针对步骤201中该各个目标可以是所有ID所属的各个目标。在该步骤202中，如果电子设备具有显示功能，则可以在用户交互界面上展示该各个目标的局部图像，如从某一帧视频图像中截取的包含该目标的局部图像。如果电子设备没有显示功能，则可以将该各个目标的局部图像发送给用户设备，由用户设备在用户交互界面上展示该各个目标的局部图像。
当视频分析结束后，电子设备与用户进行交互，将该视频中所有出现的目标都通过用户交互界面展示给用户，用户可以通过浏览选择需要遮挡或者不需要遮挡的目标，电子设备可以根据用户的选择确定需要遮挡的第一目标或者不需要遮挡的第二目标，该第一目标是用户不想关注或不感兴趣的无关目标，第二目标是用户希望关注或感兴趣的目标。
203、确定该各个目标中与该第一目标为同一个真实目标的目标。
针对步骤201中当同一目标多次进出监控区域时，电子设备会认为该目标为多个目标的情况，该各个目标中的有些目标可能为同一个真实目标。
因此，对于需要进行遮挡的第一目标，为了保证遮挡的全面性，电子设备可以通过目标对比，从该各个目标中找到与该第一目标为同一个真实目标的目标。其中，目标对比可以是将两个目标的图像特征利用预设计算方法，得到它们之间的相似度。在一种可能实现方式中，电子设备可以通过各个目标与第一目标的相似度，来确定与该第一目标为同一个真实目标的目标。具体地，电子设备可以将该第一目标的图像特征和该各个目标的图像特征进行比对，获取该各个目标与该第一目标的相似度；根据该各个目标与该第一目标的相似度，确定该各个目标中与该第一目标为同一个真实目标的目标。
当用户选定第一目标后，电子设备可以根据第一目标的ID，获取与该ID对应存储的该第一目标的一个或多个图像特征，然后将视频中所有目标的图像特征与该第一目标的图像特征进行两两对比，计算相似度。例如，电子设备可以利用欧式距离计算两个图像特征的相似度，欧式距离越小相似度越大。当然，相似度的计算不限于欧式距离，如还可以是余弦相似度，本公开实施例对相似度的计算方式不做具体限定。
电子设备根据该各个目标与该第一目标的相似度，确定该各个目标中与该第一目标为同一个真实目标的目标包括但不限于以下两种可能实现方式：
第一种方式、根据该各个目标与该第一目标的相似度，将该各个目标进行排列展示，相似度越大排列越靠前；当检测到目标确认事件时，将该目标确认事件对应的目标确定为与该第一目标为同一个真实目标的目标，该目标确认事件用于从该各个目标中选择与该第一目标为同一个真实目标的目标。
该方式是电子设备根据用户的确认操作，来确定与该第一目标为同一个真实目标的目标。针对步骤201描述的一个真实目标存在多个ID的情况，通过将比对后的目标，按照相似度从高到低的排序展示给用户，由用户根据比对结果确认排序靠前的目标是否为同一个真实目标，并进行选择，如选择与第一目标为同一个真实目标的多个目标(多个ID)。
通过将视频分析与简单的人工确认结合，可以高效准确的找到所有需要遮挡的目标，进而高效的实现对视频中无关目标进行有效的隐私遮挡，同时对感兴趣的目标不进行遮挡。
第二种方式、根据该各个目标与该第一目标的相似度，将与该第一目标的相似度大于预设阈值的目标确定为与该第一目标为同一个真实目标的目标。
其中，该预设阈值可以由用户根据实际需求进行设置，或者，也可以由该电子设备默认设置，本公开实施例对此不做限定。
该方式是电子设备根据各个目标与该第一目标的相似度大小，来确定与该第一目标为同一个真实目标的目标，可以减少用户操作。当任一目标与该第一目标的相似度大于预设阈值时，电子设备可以认为该目标与该第一目标为同一个真实目标，均是需要遮挡的目标。
204、根据该目标和该第一目标在该多帧视频图像中的位置信息和尺寸信息，对该多帧视频图像中该目标和该第一目标所在的区域进行遮挡。
其中，该目标所在的区域为该目标的位置信息和尺寸信息所对应的区域。
本公开实施例中，电子设备在确定与需要遮挡的第一目标为同一个真实目标的目标后，可以获取已存储的第一目标以及与第一目标为同一个真实目标的所有目标的运动轨迹信息，从而进行隐私遮挡操作。隐私遮挡是指在图片中或视频中采用某种技术手段对敏感的目标进行遮挡。
如电子设备可以在该多帧图像中的每帧图像中，将与第一目标为同一个真实目标的所有目标进行遮挡，从而生成遮挡视频，遮挡视频是指对原视频中的特定目标进行遮挡后得到的视频。具体地，对于每个需要遮挡的目标，电子设备可以根据该目标在每帧视频图像中的位置信息和尺寸信息，在每帧视频图像 中对该目标进行遮挡。其中，遮挡方式包括但不限于在相应区域叠加不透明的遮挡块，这个遮挡块可以设置为目标颜色或马赛克形式，只要能起到遮挡作用即可。
需要说明的是，该步骤203和步骤204是根据该第一目标的运动轨迹信息，对该多帧视频图像中该第一目标所在的区域进行遮挡的一种可能实现方式，该第一目标所在的区域为该第一目标的位置信息和尺寸信息所对应的区域。通过将与第一目标为同一个真实目标的所有目标进行遮挡，可以保证遮挡的全面性。在另一实施例中，根据该第一目标的运动轨迹信息，对该多帧视频图像中该第一目标所在的区域进行遮挡还可以包括其他可能的实现方式，譬如，可以直接根据该第一目标的运动轨迹信息在该多帧视频图像中确定需要遮挡的区域，然后在该多帧视频图像中对所确定的区域进行遮挡。
参见图3，图3是本公开实施例提供的一种生成遮挡视频的流程示意图。如图3所示，整个过程可以由数据提取单元、数据存储单元、遮挡处理单元和用户交互单元来实现，其中，数据提取单元包括对视频进行目标检测、目标跟踪、目标评价和目标特征提取。数据提取单元负责提取给定视频中的所有目标，对其进行跟踪，跟踪的每个目标生成唯一标识ID，在跟踪过程中获取每个目标在时序上的轨迹，譬如，获取每个目标在每帧视频图像中的位置和尺寸。在目标跟踪过程中，对目标进行实时评价，示例性地，进行实时评价的评价指标可以包含但不限于姿态、尺寸、成像条件、遮挡因素、角度等，并根据目标类型挑选满足评价条件的一个或多个目标来提取图像特征。数据存储单元可以采用元数据(Metadata)的形式进行数据存储，将数据提取单元提取出来的每个目标的ID、运动轨迹信息和图像特征存储起来。遮挡处理单元结合用户交互单元提供半自动的隐私遮挡操作，包括提供目标预览，由用户选择关注目标，获取该目标的图像特征，进行特征比对，人工对比对结果进行复核筛选，然后获取已存储的该目标的运动轨迹信息，最后生成遮挡视频。
需要说明的是，本公开实施例是以步骤201至步骤204以电子设备执行为例进行说明，也即是，该电子设备上可以集成有图3中数据提取单元、数据存储单元、遮挡处理单元和用户交互单元所实现的各个功能。当然，该步骤201至步骤204也可以由不同的设备执行，也即是，图3中的各个单元所实现的各个功能可以分别集成在不同的设备上。
通过利用图像特征比对来确定与需要遮挡的目标为同一个真实目标的所有 目标，可以实现对需要遮挡的所有目标进行遮挡，对不需要遮挡的所有目标不进行遮挡的功能。同时，还可以实现同一个真实目标多次进出同一场景的简易遮挡方法，该方法还可以用于不同场景下对同一目标的遮挡。例如，如该不同场景可以是同一监控区域内不同摄像头拍摄到不同视频的场景，针对不同场景的不同视频均可以通过上述步骤201至步骤204来实现对该监控区域内同一目标的遮挡。相比于相关技术中人工对视频中需要遮挡的目标逐帧添加坐标信息，最终根据每帧的坐标信息进行遮挡，本公开实施例提供的技术方案在处理视频量较大的情况下，可以大幅降低用户的人工操作成本。
在本公开实施例中，获取视频内出现的各个目标的运动轨迹信息，由于目标的运动轨迹信息可以包括目标在多帧视频图像中的位置信息和尺寸信息，因而在确定需要遮挡的第一目标后，可以根据第一目标的运动轨迹信息，对该多帧视频图像中该第一目标所在区域进行遮挡，每帧视频图像中该第一目标所在的区域可以不同，从而实现对需要遮挡的目标进行准确有效的遮挡。
图4是本公开实施例提供的一种视频图像遮挡装置的结构示意图。参照图4，该装置包括：
获取模块401，用于基于视频的多帧视频图像，获取该视频内出现的各个目标的运动轨迹信息，该各个目标的运动轨迹信息包括该各个目标在该多帧视频图像中的位置信息和尺寸信息；
确定模块402，用于确定该各个目标中需要遮挡的第一目标；
遮挡模块403，用于根据该第一目标的运动轨迹信息，对该多帧视频图像中该第一目标所在的区域进行遮挡，该第一目标所在的区域为该第一目标的位置信息和尺寸信息所对应的区域。
在一种可能实现方式中，该获取模块用于对于该多帧视频图像中的每帧视频图像，对该视频图像进行目标检测，确定该视频图像中的多个目标；获取该多个目标在该视频图像中的位置信息和尺寸信息。
在一种可能实现方式中，参见图5，该装置还包括：
第一生成模块404，用于当该视频图像为该多帧视频图像中的第一帧视频图像时，生成该多个目标中每个目标的唯一标识；
第一存储模块405，用于将该多个目标中每个目标在该视频图像中的位置信息和尺寸信息与该每个目标的唯一标识对应存储。
在一种可能实现方式中，参见图6，该装置还包括：
该确定模块402还用于当该视频图像为该多帧视频图像中除第一帧视频图像以外的视频图像时，确定该多个目标中的已知目标和未知目标，该已知目标为该视频图像的上一帧视频图像中包含的目标，该未知目标为该上一帧视频图像中未包含的目标；
第二生成模块406，用于生成该未知目标的唯一标识；
第二存储模块407，用于将该已知目标在该视频图像中的位置信息和尺寸信息与该已知目标的唯一标识对应存储；将该未知目标在该视频图像中的位置信息和尺寸信息与该未知目标的唯一标识对应存储。
在一种可能实现方式中，参见图7，该装置还包括：
提取模块408，用于提取该视频图像中该多个目标的图像特征；
第三存储模块409，用于将该多个目标的图像特征与该多个目标的唯一标识对应存储。
在一种可能实现方式中，该提取模块408用于获取该视频图像中该多个目标的评价信息，该评价信息包括姿态、尺寸、成像条件、遮挡情况和拍摄角度中至少一项；从该多个目标中，选取评价信息满足预设评价条件的目标；提取所选取的目标的图像特征。
在一种可能实现方式中，该确定模块402用于将该各个目标进行展示；当检测到第一选择事件时，将该第一选择事件对应的目标确定为该第一目标，该第一选择事件用于从该各个目标中选择需要遮挡的目标；当检测到第二选择事件时，将除该第二选择事件对应的目标以外的目标确定为该第一目标，该第二选择事件用于从该各个目标中选择不需要遮挡的目标。
在一种可能实现方式中，该遮挡模块403用于确定该各个目标中与该第一目标为同一个真实目标的目标；根据该目标和该第一目标在该多帧视频图像中的位置信息和尺寸信息，对该多帧视频图像中该目标和该第一目标所在的区域进行遮挡。
在一种可能实现方式中，该确定模块402用于将该第一目标的图像特征和该各个目标的图像特征进行比对，获取该各个目标与该第一目标的相似度；根据该各个目标与该第一目标的相似度，确定该各个目标中与该第一目标为同一个真实目标的目标。
在一种可能实现方式中，该确定模块402用于根据该各个目标与该第一目 标的相似度，将该各个目标进行排列展示，相似度越大排列越靠前；当检测到目标确认事件时，将该目标确认事件对应的目标确定为与该第一目标为同一个真实目标的目标，该目标确认事件用于从该各个目标中选择与该第一目标为同一个真实目标的目标。
在一种可能实现方式中，该确定模块402用于根据该各个目标与该第一目标的相似度，将与该第一目标的相似度大于预设阈值的目标确定为与该第一目标为同一个真实目标的目标。
在本公开实施例中，获取视频内出现的各个目标的运动轨迹信息，由于目标的运动轨迹信息可以包括目标在多帧视频图像中的位置信息和尺寸信息，因而在确定需要遮挡的第一目标后，可以根据第一目标的运动轨迹信息，对该多帧视频图像中该第一目标所在区域进行遮挡，每帧视频图像中该第一目标所在的区域可以不同，从而实现对需要遮挡的目标进行准确有效的遮挡。
需要说明的是：上述实施例提供的视频图像遮挡装置在视频图像遮挡时，仅以上述各功能模块的划分进行举例说明，实际应用中，可以根据需要而将上述功能分配由不同的功能模块完成，即将设备的内部结构划分成不同的功能模块，以完成以上描述的全部或者部分功能。另外，上述实施例提供的视频图像遮挡装置与视频图像遮挡方法实施例属于同一构思，其具体实现过程详见方法实施例，这里不再赘述。
图8是本公开实施例提供的一种电子设备800的结构示意图，该电子设备800可因配置或性能不同而产生比较大的差异，可以包括一个或一个以上处理器(Central Processing Units，CPU)801和一个或一个以上的存储器802，其中，该存储器802中存储有至少一条指令，该至少一条指令由该处理器801加载并执行以实现上述各个方法实施例提供的视频图像遮挡方法。当然，该电子设备800还可以具有有线或无线网络接口、键盘以及输入输出接口等部件，以便进行输入输出，该电子设备800还可以包括其他用于实现设备功能的部件，在此不做赘述。
在示例性实施例中，还提供了一种存储有至少一条指令的计算机可读存储介质，例如存储有至少一条指令的存储器，上述至少一条指令被处理器执行时 实现上述实施例中的视频图像遮挡方法。例如，该计算机可读存储介质可以是只读内存(Read-Only Memory,ROM)、随机存取存储器(Random Access Memory，RAM)、只读光盘(Compact Disc Read-Only Memory，CD-ROM)、磁带、软盘和光数据存储设备等。
本领域普通技术人员可以理解实现上述实施例的全部或部分步骤可以通过硬件来完成，也可以通过程序来指令相关的硬件完成，上述程序可以存储于一种计算机可读存储介质中，上述提到的存储介质可以是只读存储器，磁盘或光盘等。
上述仅为本公开的较佳实施例，并不用以限制本公开，凡在本公开的精神和原则之内，所作的任何修改、等同替换、改进等，均应包含在本公开的保护范围之内。

Claims (24)

  1. 一种视频图像遮挡方法，其特征在于，所述方法包括：
    基于视频的多帧视频图像，获取所述视频内出现的各个目标的运动轨迹信息，所述各个目标的运动轨迹信息包括所述各个目标在所述多帧视频图像中的位置信息和尺寸信息；
    确定所述各个目标中需要遮挡的第一目标；
    根据所述第一目标的运动轨迹信息，对所述多帧视频图像中所述第一目标所在的区域进行遮挡，所述第一目标所在的区域为所述第一目标的位置信息和尺寸信息所对应的区域。
  2. 根据权利要求1所述的方法，其特征在于，所述基于视频的多帧视频图像，获取所述视频内出现的各个目标的运动轨迹信息，包括：
    对于所述多帧视频图像中的每帧视频图像，对所述视频图像进行目标检测，确定所述视频图像中的多个目标；
    获取所述多个目标在所述视频图像中的位置信息和尺寸信息。
  3. 根据权利要求2所述的方法，其特征在于，所述对所述视频图像进行目标检测，确定所述视频图像中的多个目标之后，所述方法还包括：
    当所述视频图像为所述多帧视频图像中的第一帧视频图像时，生成所述多个目标中每个目标的唯一标识；
    所述获取所述多个目标在所述视频图像中的位置信息和尺寸信息之后，所述方法还包括：
    将所述多个目标中每个目标在所述视频图像中的位置信息和尺寸信息与所述每个目标的唯一标识对应存储。
  4. 根据权利要求2所述的方法，其特征在于，所述对所述视频图像进行目标检测，确定所述视频图像中的多个目标之后，所述方法还包括：
    当所述视频图像为所述多帧视频图像中除第一帧视频图像以外的视频图像时，确定所述多个目标中的已知目标和未知目标，所述已知目标为所述视频图像的上一帧视频图像中包含的目标，所述未知目标为所述上一帧视频图像中未 包含的目标；
    生成所述未知目标的唯一标识；
    所述获取所述多个目标在所述视频图像中的位置信息和尺寸信息之后，所述方法还包括：
    将所述已知目标在所述视频图像中的位置信息和尺寸信息与所述已知目标的唯一标识对应存储；
    将所述未知目标在所述视频图像中的位置信息和尺寸信息与所述未知目标的唯一标识对应存储。
  5. 根据权利要求3或4所述的方法，其特征在于，所述对所述视频图像进行目标检测，确定所述视频图像中的多个目标之后，所述方法还包括：
    提取所述视频图像中所述多个目标的图像特征；
    将所述多个目标的图像特征与所述多个目标的唯一标识对应存储。
  6. 根据权利要求5所述的方法，其特征在于，所述提取所述视频图像中所述多个目标的图像特征，包括：
    获取所述视频图像中所述多个目标的评价信息，所述评价信息包括姿态、尺寸、成像条件、遮挡情况和拍摄角度中至少一项；
    从所述多个目标中，选取评价信息满足预设评价条件的目标；
    提取所选取的目标的图像特征。
  7. 根据权利要求1所述的方法，其特征在于，所述确定所述各个目标中需要遮挡的第一目标，包括：
    将所述各个目标进行展示；
    当检测到第一选择事件时，将所述第一选择事件对应的目标确定为所述第一目标，所述第一选择事件用于从所述各个目标中选择需要遮挡的目标；
    当检测到第二选择事件时，将除所述第二选择事件对应的目标以外的目标确定为所述第一目标，所述第二选择事件用于从所述各个目标中选择不需要遮挡的目标。
  8. 根据权利要求1或7所述的方法，其特征在于，所述根据所述第一目标的运动轨迹信息，对所述多帧视频图像中所述第一目标所在的区域进行遮挡，包括：
    确定所述各个目标中与所述第一目标为同一个真实目标的目标；
    根据所述目标和所述第一目标在所述多帧视频图像中的位置信息和尺寸信息，对所述多帧视频图像中所述目标和所述第一目标所在的区域进行遮挡。
  9. 根据权利要求8所述的方法，其特征在于，所述确定所述各个目标中与所述第一目标为同一个真实目标的目标，包括：
    将所述第一目标的图像特征和所述各个目标的图像特征进行比对，获取所述各个目标与所述第一目标的相似度；
    根据所述各个目标与所述第一目标的相似度，确定所述各个目标中与所述第一目标为同一个真实目标的目标。
  10. 根据权利要求9所述的方法，其特征在于，所述根据所述各个目标与所述第一目标的相似度，确定所述各个目标中与所述第一目标为同一个真实目标的目标，包括：
    根据所述各个目标与所述第一目标的相似度，将所述各个目标进行排列展示，相似度越大排列越靠前；
    当检测到目标确认事件时，将所述目标确认事件对应的目标确定为与所述第一目标为同一个真实目标的目标，所述目标确认事件用于从所述各个目标中选择与所述第一目标为同一个真实目标的目标。
  11. 根据权利要求9所述的方法，其特征在于，所述根据所述各个目标与所述第一目标的相似度，确定所述各个目标中与所述第一目标为同一个真实目标的目标，包括：
    根据所述各个目标与所述第一目标的相似度，将与所述第一目标的相似度大于预设阈值的目标确定为与所述第一目标为同一个真实目标的目标。
  12. 一种视频图像遮挡装置，其特征在于，所述装置包括：
    获取模块，用于基于视频的多帧视频图像，获取所述视频内出现的各个目标的运动轨迹信息，所述各个目标的运动轨迹信息包括所述各个目标在所述多帧视频图像中的位置信息和尺寸信息；
    确定模块，用于确定所述各个目标中需要遮挡的第一目标；
    遮挡模块，用于根据所述第一目标的运动轨迹信息，对所述多帧视频图像中所述第一目标所在的区域进行遮挡，所述第一目标所在的区域为所述第一目标的位置信息和尺寸信息所对应的区域。
  13. 根据权利要求12所述的装置，其特征在于，所述获取模块用于对于所述多帧视频图像中的每帧视频图像，对所述视频图像进行目标检测，确定所述视频图像中的多个目标；获取所述多个目标在所述视频图像中的位置信息和尺寸信息。
  14. 根据权利要求13所述的装置，其特征在于，所述装置还包括：
    第一生成模块，用于当所述视频图像为所述多帧视频图像中的第一帧视频图像时，生成所述多个目标中每个目标的唯一标识；
    第一存储模块，用于将所述多个目标中每个目标在所述视频图像中的位置信息和尺寸信息与所述每个目标的唯一标识对应存储。
  15. 根据权利要求13所述的装置，其特征在于，所述装置还包括：
    所述确定模块还用于当所述视频图像为所述多帧视频图像中除第一帧视频图像以外的视频图像时，确定所述多个目标中的已知目标和未知目标，所述已知目标为所述视频图像的上一帧视频图像中包含的目标，所述未知目标为所述上一帧视频图像中未包含的目标；
    第二生成模块，用于生成所述未知目标的唯一标识；
    第二存储模块，用于将所述已知目标在所述视频图像中的位置信息和尺寸信息与所述已知目标的唯一标识对应存储；将所述未知目标在所述视频图像中的位置信息和尺寸信息与所述未知目标的唯一标识对应存储。
  16. 根据权利要求14或15所述的装置，其特征在于，所述装置还包括：
    提取模块，用于提取所述视频图像中所述多个目标的图像特征；
    第三存储模块，用于将所述多个目标的图像特征与所述多个目标的唯一标识对应存储。
  17. 根据权利要求16所述的装置，其特征在于，所述提取模块用于获取所述视频图像中所述多个目标的评价信息，所述评价信息包括姿态、尺寸、成像条件、遮挡情况和拍摄角度中至少一项；从所述多个目标中，选取评价信息满足预设评价条件的目标；提取所选取的目标的图像特征。
  18. 根据权利要求12所述的装置，其特征在于，所述确定模块用于将所述各个目标进行展示；当检测到第一选择事件时，将所述第一选择事件对应的目标确定为所述第一目标，所述第一选择事件用于从所述各个目标中选择需要遮挡的目标；当检测到第二选择事件时，将除所述第二选择事件对应的目标以外的目标确定为所述第一目标，所述第二选择事件用于从所述各个目标中选择不需要遮挡的目标。
  19. 根据权利要求12或18所述的装置，其特征在于，所述遮挡模块用于确定所述各个目标中与所述第一目标为同一个真实目标的目标；根据所述目标和所述第一目标在所述多帧视频图像中的位置信息和尺寸信息，对所述多帧视频图像中所述目标和所述第一目标所在的区域进行遮挡。
  20. 根据权利要求19所述的装置，其特征在于，所述确定模块用于将所述第一目标的图像特征和所述各个目标的图像特征进行比对，获取所述各个目标与所述第一目标的相似度；根据所述各个目标与所述第一目标的相似度，确定所述各个目标中与所述第一目标为同一个真实目标的目标。
  21. 根据权利要求20所述的装置，其特征在于，所述确定模块用于根据所述各个目标与所述第一目标的相似度，将所述各个目标进行排列展示，相似度越大排列越靠前；当检测到目标确认事件时，将所述目标确认事件对应的目标确定为与所述第一目标为同一个真实目标的目标，所述目标确认事件用于从所 述各个目标中选择与所述第一目标为同一个真实目标的目标。
  22. 根据权利要求20所述的装置，其特征在于，所述确定模块用于根据所述各个目标与所述第一目标的相似度，将与所述第一目标的相似度大于预设阈值的目标确定为与所述第一目标为同一个真实目标的目标。
  23. 一种电子设备，其特征在于，包括处理器和存储器；所述存储器，用于存放至少一条指令；所述处理器，用于执行所述存储器上所存放的至少一条指令：
    基于视频的多帧视频图像，获取所述视频内出现的各个目标的运动轨迹信息，所述各个目标的运动轨迹信息包括所述各个目标在所述多帧视频图像中的位置信息和尺寸信息；
    确定所述各个目标中需要遮挡的第一目标；
    根据所述第一目标的运动轨迹信息，对所述多帧视频图像中所述第一目标所在的区域进行遮挡，所述第一目标所在的区域为所述第一目标的位置信息和尺寸信息所对应的区域。
  24. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质内存储有至少一条指令，所述至少一条指令被处理器执行：
    基于视频的多帧视频图像，获取所述视频内出现的各个目标的运动轨迹信息，所述各个目标的运动轨迹信息包括所述各个目标在所述多帧视频图像中的位置信息和尺寸信息；
    确定所述各个目标中需要遮挡的第一目标；
    根据所述第一目标的运动轨迹信息，对所述多帧视频图像中所述第一目标所在的区域进行遮挡，所述第一目标所在的区域为所述第一目标的位置信息和尺寸信息所对应的区域。
PCT/CN2019/121644 2018-11-28 2019-11-28 视频图像遮挡方法、装置、设备及存储介质 WO2020108573A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201811435929.1A CN111241872B (zh) 2018-11-28 2018-11-28 视频图像遮挡方法及装置
CN201811435929.1 2018-11-28

Publications (1)

Publication Number Publication Date
WO2020108573A1 true WO2020108573A1 (zh) 2020-06-04

Family

ID=70854339

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/121644 WO2020108573A1 (zh) 2018-11-28 2019-11-28 视频图像遮挡方法、装置、设备及存储介质

Country Status (2)

Country Link
CN (1) CN111241872B (zh)
WO (1) WO2020108573A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113807445A (zh) * 2021-09-23 2021-12-17 城云科技（中国）有限公司 案卷复核方法、装置及电子装置、计算机程序产品

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111654700B (zh) * 2020-06-19 2022-12-06 杭州海康威视数字技术股份有限公司 一种隐私遮蔽处理方法、装置、电子设备及监控系统
CN111985419B (zh) * 2020-08-25 2022-10-14 腾讯科技（深圳）有限公司 视频处理方法及相关设备
CN112188058A (zh) * 2020-09-29 2021-01-05 努比亚技术有限公司 一种视频拍摄方法、移动终端以及计算机存储介质
CN115546900B (zh) * 2022-11-25 2023-03-31 浙江莲荷科技有限公司 风险识别方法、装置、设备及存储介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060158527A1 (en) * 2004-12-15 2006-07-20 Kang Kyun H Method and apparatus for controlling privacy mask display
CN105007395A (zh) * 2015-07-22 2015-10-28 深圳市万姓宗祠网络科技股份有限公司 一种连续记录视频、影像的隐私处理方法
CN105957001A (zh) * 2016-04-18 2016-09-21 深圳感官密码科技有限公司 一种隐私保护方法及装置
CN107820041A (zh) * 2016-09-13 2018-03-20 华为数字技术（苏州）有限公司 隐私遮挡方法及装置

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP6336872B2 (ja) * 2014-09-29 2018-06-06 Kddi株式会社 オブジェクト追跡方法、装置およびプログラム
JP6474126B2 (ja) * 2015-02-27 2019-02-27 Kddi株式会社 オブジェクト追跡方法、装置およびプログラム
EP3378227A4 (en) * 2015-11-18 2019-07-03 Jorg Tilkin PRIVACY PROTECTION IN VIDEO SURVEILLANCE SYSTEMS
CN106162091A (zh) * 2016-07-28 2016-11-23 乐视控股（北京）有限公司 一种视频监控方法及装置
CN106358069A (zh) * 2016-10-31 2017-01-25 维沃移动通信有限公司 一种视频数据处理方法及移动终端
CN107240120B (zh) * 2017-04-18 2019-12-17 上海体育学院 视频中运动目标的跟踪方法及装置
CN107564034A (zh) * 2017-07-27 2018-01-09 华南理工大学 一种监控视频中多目标的行人检测与跟踪方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060158527A1 (en) * 2004-12-15 2006-07-20 Kang Kyun H Method and apparatus for controlling privacy mask display
CN105007395A (zh) * 2015-07-22 2015-10-28 深圳市万姓宗祠网络科技股份有限公司 一种连续记录视频、影像的隐私处理方法
CN105957001A (zh) * 2016-04-18 2016-09-21 深圳感官密码科技有限公司 一种隐私保护方法及装置
CN107820041A (zh) * 2016-09-13 2018-03-20 华为数字技术（苏州）有限公司 隐私遮挡方法及装置

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113807445A (zh) * 2021-09-23 2021-12-17 城云科技（中国）有限公司 案卷复核方法、装置及电子装置、计算机程序产品
CN113807445B (zh) * 2021-09-23 2024-04-16 城云科技（中国）有限公司 案卷复核方法、装置及电子装置、可读存储介质

Also Published As

Publication number Publication date
CN111241872A (zh) 2020-06-05
CN111241872B (zh) 2023-09-22

Similar Documents

Publication Publication Date Title
WO2020108573A1 (zh) 视频图像遮挡方法、装置、设备及存储介质
US11594029B2 (en) Methods and systems for determining ball shot attempt location on ball court
US11810321B2 (en) Methods and systems for multiplayer tagging using artificial intelligence
US9600760B2 (en) System and method for utilizing motion fields to predict evolution in dynamic scenes
US11263446B2 (en) Method for person re-identification in closed place, system, and terminal device
US20180225852A1 (en) Apparatus and method for generating best-view image centered on object of interest in multiple camera images
US20130335635A1 (en) Video Analysis Based on Sparse Registration and Multiple Domain Tracking
AU2017272325A1 (en) System and method of generating a composite frame
CN111274928A (zh) 一种活体检测方法、装置、电子设备和存储介质
US9384400B2 (en) Method and apparatus for identifying salient events by analyzing salient video segments identified by sensor information
Chakraborty et al. A real-time trajectory-based ball detection-and-tracking framework for basketball video
JP2016163328A (ja) 情報処理装置、情報処理方法、およびプログラム
WO2023045183A1 (zh) 图像处理
Pidaparthy et al. Keep your eye on the puck: Automatic hockey videography
CN112954443A (zh) 全景视频的播放方法、装置、计算机设备和存储介质
CN112511859A (zh) 一种视频处理方法、装置和存储介质
González et al. Single object long-term tracker for smart control of a ptz camera
US10372994B2 (en) Method, system and apparatus for selecting a video frame
CN114037923A (zh) 一种目标活动热点图绘制方法、系统、设备及存储介质
Xu et al. Fast and accurate object detection using image cropping/resizing in multi-view 4K sports videos
Nieto et al. An automatic system for sports analytics in multi-camera tennis videos
AU2015258346A1 (en) Method and system of transitioning between images
Stefański et al. The problem of detecting boxers in the boxing ring
Chen et al. Multi-sensored vision for autonomous production of personalized video summaries
CN115334241B (zh) 对焦控制方法、装置、存储介质及摄像设备

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: 19889653

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19889653

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 19889653

Country of ref document: EP

Kind code of ref document: A1

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

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 16.12.2021)

122 Ep: pct application non-entry in european phase

Ref document number: 19889653

Country of ref document: EP

Kind code of ref document: A1