WO2020252974A1 - 一种针对运动状态下的多目标对象追踪方法和装置 - Google Patents

一种针对运动状态下的多目标对象追踪方法和装置 Download PDF

Info

Publication number
WO2020252974A1
WO2020252974A1 PCT/CN2019/108432 CN2019108432W WO2020252974A1 WO 2020252974 A1 WO2020252974 A1 WO 2020252974A1 CN 2019108432 W CN2019108432 W CN 2019108432W WO 2020252974 A1 WO2020252974 A1 WO 2020252974A1
Authority
WO
WIPO (PCT)
Prior art keywords
video
target object
target
adjacent
video frames
Prior art date
Application number
PCT/CN2019/108432
Other languages
English (en)
French (fr)
Inventor
吉长江
Original Assignee
北京影谱科技股份有限公司
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 北京影谱科技股份有限公司 filed Critical 北京影谱科技股份有限公司
Priority to US17/620,119 priority Critical patent/US20220215560A1/en
Publication of WO2020252974A1 publication Critical patent/WO2020252974A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
    • G06T7/248—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments involving reference images or patches
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/285—Analysis of motion using a sequence of stereo image pairs
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/70—Determining position or orientation of objects or cameras
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/90—Determination of colour characteristics
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10016—Video; Image sequence
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10024—Color image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30196—Human being; Person
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30221—Sports video; Sports image

Definitions

  • the embodiments of the present application relate to the field of artificial intelligence technology, and in particular to a method and device for tracking multiple target objects in a motion state.
  • Computer vision technology is a technology that studies how to make machines "see”. Cameras and computer equipment can be used to replace human eyes with machine vision processing technologies such as real-time recognition, positioning, tracking, and measurement of target objects.
  • the image is analyzed and processed by computer equipment, so that the data obtained by the camera is more suitable for human observation or image information sent to the instrument for detection.
  • a camera it is usually necessary to use a camera to track multiple players on the court at the same time, so that the user can switch to the player's corresponding tracking shooting angle or obtain the player's movement track data on the court at any time according to needs. Therefore, how to achieve rapid and precise positioning and tracking of the target object when the video capture device and the target object are both in motion, has become a technical problem to be solved urgently.
  • the technical means usually used in the prior art is to judge the position similarity of target objects in video frames based on 2D image recognition technology, and determine whether the target objects in adjacent video frames are the same target object. Furthermore, the positioning and tracking of the target object and the movement track of the target object are obtained.
  • the technical means usually used in the prior art is to judge the position similarity of target objects in video frames based on 2D image recognition technology, and determine whether the target objects in adjacent video frames are the same target object. Furthermore, the positioning and tracking of the target object and the movement track of the target object are obtained.
  • there are often changes in the pose of the video capture device itself resulting in poor actual tracking and shooting effects of the target object in the prior art, and identification errors are prone to occur. Meet the needs of current users.
  • the embodiments of the present application provide a tracking method for multiple target objects in a motion state, so as to solve the problems of low efficiency and poor accuracy in the recognition and tracking of multiple target objects in a video in the prior art.
  • a tracking method for multiple target objects in a motion state including: obtaining video frames included in video data collected in a video capturing device; sending the video frames to a preset feature recognition model , Determine the feature detection area corresponding to the target object in the video frame, extract the color feature of the target object from the detection area, and compare the color feature of the target object in the adjacent video frames , Obtain the first comparison result; determine the position information of the identification part of the target object in the adjacent video frame in the target coordinate system, and place the identification part in the adjacent video frame in the target coordinate system To compare the position information of to obtain a second comparison result; according to the first comparison result and the second comparison result, determine whether the target object in the adjacent video frame is the same target object, If so, the target object in the adjacent video frames is taken as the same target object for tracking.
  • the determining the position information of the identification part of the target object in the adjacent video frames in the target coordinate system specifically includes: predicting the positions of the video acquisition devices corresponding to the adjacent video frames.
  • the pose change information obtain the pose change information of the video acquisition device corresponding to the adjacent video frames; according to the pose change information and the video corresponding to the previous video frame in the adjacent video frames
  • the location information of the collection device is determined, and the location information of the video collection device corresponding to the next video frame in the adjacent video frames is determined;
  • the identification position of the target object is obtained by triangulation method to obtain the position information of the identification position of the target object in the spatial rectangular coordinate system constructed with the spatial coordinate origin of the video capture device; the position information of the target object is obtained through coordinate transformation The position information of the identified part in the target coordinate system.
  • the method for tracking multiple target objects in a motion state further includes: determining the actual motion area of the target object in the video frame; taking the actual motion area of the target object in the video frame as For the area to be detected, the feature detection area outside the area to be detected is filtered out to obtain the feature detection area within the area to be detected.
  • the identification part is the neck part of the target object; correspondingly, the position information of the identification part of the target object in the target coordinate system is that the neck part of the target object is located in the to-be-detected
  • the center of the area is the position information in the space rectangular coordinate system constructed by the origin of the space coordinates.
  • the video data collected in the video acquisition device is obtained, and the video data is segmented to obtain the video segments included in the video data; the feature similarity between the video segments is detected, and the A video clip whose feature similarity reaches or exceeds a preset similarity threshold and a time interval does not exceed the preset time threshold is taken as a video shot; and the video frames included in the video shot are acquired.
  • an embodiment of the present application also provides a tracking device for multiple target objects in a moving state, including: a video frame obtaining unit, configured to obtain video frames included in the video data collected in the video capture device; first comparison Unit, configured to send the video frame to a preset feature recognition model, determine the feature detection area corresponding to the target object in the video frame, extract the color feature of the target object from the detection area, and compare the corresponding The color features of the target object in the adjacent video frames are compared to obtain a first comparison result; a second comparison unit is used to determine that the identification position of the target object in the adjacent video frames is The position information in the target coordinate system is used to compare the position information of the identification parts in the adjacent video frames in the target coordinate system to obtain a second comparison result; The comparison result and the second comparison result determine whether the target objects in the adjacent video frames are the same target object, and if so, the target objects in the adjacent video frames are regarded as the same target object follow it up.
  • the determining the position information of the identification part of the target object in the adjacent video frames in the target coordinate system specifically includes: predicting the positions of the video acquisition devices corresponding to the adjacent video frames.
  • the pose change information obtain the pose change information of the video acquisition device corresponding to the adjacent video frames; according to the pose change information and the video corresponding to the previous video frame in the adjacent video frames
  • the location information of the collection device is determined, and the location information of the video collection device corresponding to the next video frame in the adjacent video frames is determined;
  • the identification position of the target object is obtained by triangulation method to obtain the position information of the identification position of the target object in the spatial rectangular coordinate system constructed with the spatial coordinate origin of the video capture device; the position information of the target object is obtained through coordinate transformation The position information of the identified part in the target coordinate system.
  • the device for tracking multiple target objects in a motion state further includes: a motion area determining unit, configured to determine the actual motion area of the target object in the video frame; and a filtering unit, configured to compare all The actual motion area of the target object in the video frame is used as the area to be detected, and the feature detection area outside the area to be detected is filtered out to obtain the feature detection area within the area to be detected.
  • the identification part is the neck part of the target object; correspondingly, the position information of the identification part of the target object in the target coordinate system is that the neck part of the target object is located in the to-be-detected
  • the center of the area is the position information in the space rectangular coordinate system constructed by the origin of the space coordinates.
  • the obtaining the video frames included in the video data collected in the video collecting device specifically includes: obtaining the video data collected in the video collecting device, and performing segmentation processing on the video data to obtain the video Video clips included in the data; detecting the feature similarity between the video clips, and using the video clips whose feature similarity reaches or exceeds a preset similarity threshold and the time interval does not exceed the preset time threshold as a video shot; Obtain the video frames included in the video shot.
  • the present application also provides an electronic device, including: a processor and a memory; wherein the memory is used to store a program for the tracking method for multiple target objects in a motion state, and the device is powered on and runs through the processor After the procedure for the multi-target tracking method in motion, the following steps are performed:
  • the position information of the marking part in the target coordinate system is compared with the position information of the marking part in the adjacent video frames in the target coordinate system to obtain a second comparison result; according to the first comparison result and
  • the second comparison result determines whether the target object in the adjacent video frame is the same target object, and if so, the target object in the adjacent video frame is tracked as the same target object.
  • the present application also provides a storage device that stores a program for the multi-target object tracking method in a motion state, and the program is run by the processor to perform the following steps:
  • the position information of the marking part in the target coordinate system is compared with the position information of the marking part in the adjacent video frames in the target coordinate system to obtain a second comparison result; according to the first comparison result and
  • the second comparison result determines whether the target object in the adjacent video frame is the same target object, and if so, the target object in the adjacent video frame is tracked as the same target object.
  • Using the method for tracking multiple target objects in motion as described in this application can simultaneously quickly identify and track multiple target objects in motion, which improves the recognition of multiple target objects in motion in video data. And tracking accuracy, thereby enhancing the user experience.
  • FIG. 1 is a flowchart of a method for tracking multiple target objects in a motion state according to an embodiment of the application
  • FIG. 2 is a schematic diagram of a tracking device for multiple target objects in a motion state provided by an embodiment of the application
  • FIG. 3 is a schematic diagram of locating a target object using a triangulation method according to an embodiment of the application
  • FIG. 4 is a schematic diagram of an electronic device provided by an embodiment of the application.
  • FIG. 1 it is a flowchart of a method for tracking multiple targets in a motion state provided by an embodiment of this application.
  • the specific implementation process includes the following steps:
  • Step S101 Obtain video frames included in the video data collected in the video collection device.
  • the video acquisition device includes video data acquisition equipment such as cameras, video recorders, and image sensors.
  • the video data is video data contained in an independent shot.
  • an independent lens is video data obtained by a continuous shooting process of the video capture device, and the video data is composed of video frames, and a group of continuous video frames can form a lens.
  • a complete video data may include multiple shots, and the acquisition of the video frames included in the video data collected in the video collection device can be specifically implemented in the following ways:
  • Video frames To obtain the video data collected in the video capture device, before obtaining the video frame contained in one of the shots, it is necessary to first perform shot segmentation on the complete video data based on the global and local features of the video frame to obtain a series of independent Video clips. The similarity between the video clips is detected, and the video clips whose similarity reaches or exceeds the preset similarity threshold and the time interval does not exceed the preset time threshold are taken as a video shot, and then the video shots contained in the video shots are obtained Video frames.
  • the color characteristics of the video frames contained in different shots usually have obvious differences.
  • the color feature extraction algorithm can extract the RGB or HSV color histogram of each video frame in the video data, and then use the window function to calculate the probability distribution of the first half and the second half of the video frame. If the two probabilities are different, the current The center of the window is the lens boundary.
  • the shot segmentation of the complete video data based on the global features and local features of the video frame can be specifically implemented through the following process:
  • Global feature analysis Calculate the first similarity between adjacent video frames of the video data based on the color features of adjacent video frames, compare the first similarity with a first similarity threshold, if the first similarity If it is less than the first similarity threshold, the video frame is regarded as a candidate video frame of an independent shot.
  • Local feature analysis Calculate the distance value between the descriptor of the key point in the candidate video frame and the previous video frame to each visual word respectively, and correspond the descriptor with the visual word with the smallest distance value of the visual word, based on Descriptors and corresponding visual words, respectively construct the visual word histograms of the candidate video frame and the previous frame, and calculate the second similarity between the visual word histograms of the video frames.
  • Shot segmentation step judge the second similarity, if the second similarity is greater than or equal to the second similarity threshold, merge the candidate video frame and the previous frame into the same shot, if all If the second similarity is less than the second similarity threshold, the candidate video frame is determined as the starting video frame of the new shot.
  • Step S102 Send the video frame to a preset feature recognition model, determine the feature detection area corresponding to the target object in the video frame, extract the color feature of the target object from the feature detection area, and compare The color features of the target object in the adjacent video frames are compared to obtain a first comparison result.
  • step S101 After obtaining the video frames included in the video data collected by the video capture device in the above step S101, data preparation is done for this step to compare the color features of the target object in the adjacent video frames.
  • step S102 the color feature of the target object may be extracted from the video frame, and the color feature of the target object in the adjacent video frames may be further compared to obtain a first comparison result.
  • the feature recognition model may refer to a Faster RCNN deep neural network model obtained through iterative training in advance.
  • the feature detection area may refer to the detection frame corresponding to each target object in the video frame obtained in the process of using the Faster RCNN deep neural network model to detect the target object on the video frame.
  • RGB red, green, blue
  • HSV Human Saturation Value, hue, saturation, lightness
  • the feature detection area corresponding to the target object in the video frame there may be a detection area generated corresponding to the non-target object in the final detection result (that is, the monitoring frame corresponding to the non-target object), At this time, the above detection results need to be filtered in advance, and only the feature detection area corresponding to the target object (that is, the monitoring frame corresponding to the target object) is retained.
  • the specific implementation is as follows:
  • the actual motion area is the movement area of the target object.
  • the difference between the color features of the stadium floor and the color features of the audience can be used to differentiate and filter through the threshold filtering method to obtain an image that only contains the stadium, and a series of processing operations such as corrosion and expansion of the stadium image are further performed.
  • Obtain the outer contour of the court (the area enclosed by the outer contour is the actual movement area of the target object), filter out the detection frame outside the outer contour of the court, and only keep the area enclosed by the outer contour ( Namely: stay in the field) the detection frame.
  • Step S103 Determine the position information of the identification part of the target object in the adjacent video frame in the target coordinate system, and compare the position information of the identification part in the adjacent video frame in the target coordinate system Yes, get the second comparison result.
  • this step can further determine the position information of the identification part of the target object in the adjacent video frames in the target coordinate system, and combine all the adjacent video frames The position information of the marking part in the target coordinate system is compared to obtain a second comparison result.
  • the target coordinate system may refer to the world coordinate system, which may refer to the absolute coordinate system of the video screen, and the coordinates of the points corresponding to the identification parts of all target objects in the video screen can be
  • the world coordinate system is used to determine the specific location of each target object.
  • the world coordinate system may refer to a spatial rectangular coordinate system constructed with the center of the detection area as the origin of the spatial coordinate system.
  • point P can refer to the position of the point corresponding to the neck of the target object
  • point Q1 can refer to the position of the point corresponding to the video capture device in the previous video frame, or it can refer to the point corresponding to the video capture device The position of the point in the previous shot
  • Q2 point can refer to the position of the point corresponding to the video capture device in the next video frame relative to the previous video, or the point corresponding to the video capture device The position in the next lens relative to the previous lens.
  • the determination of the position information of the identification part of the target object in the adjacent video frame in the target coordinate system may be specifically implemented in the following manner:
  • the visual mileage calculation method can be used to predict the pose change of the video acquisition device, and the corresponding adjacent video frames can be obtained through prediction.
  • the position and posture change information of the video capture device is further obtained, and the posture change information of the video capture device corresponding to the adjacent video frames is obtained. According to the posture change information, the position information of the video capture devices corresponding to the adjacent video frames can be determined.
  • the position information of the video acquisition device in the previous video frame in the adjacent video frames may be recorded as the first position, and the position information of the video acquisition device in the next video frame in the adjacent video frames Recorded as the second position.
  • the target object can be obtained by calculation using the triangulation method shown in FIG.
  • the position information in the spatial rectangular coordinate system constructed with the video capture device as the spatial coordinate origin can be further obtained by coordinate transformation to obtain the position information of the target object in the target coordinate system (ie, the world coordinate system).
  • the posture changes include changes in motion trajectories and activity postures.
  • the identification part may select the neck part of the target object.
  • the position information of the identification part of the target object in the target coordinate system is the position information of the neck part in the spatial rectangular coordinate system constructed with the center of the area to be detected as the spatial coordinate origin.
  • a bone detection algorithm can be used to obtain the point P corresponding to the neck of each target object.
  • Step S104 According to the first comparison result and the second comparison result, determine whether the target objects in the adjacent video frames are the same target object, and if so, compare the The target object is tracked as the same target object.
  • this step can determine all the adjacent video frames based on the first comparison result and the second comparison result. Whether the target object is the same target object, so as to realize real-time positioning and tracking of the target object.
  • the target object in the adjacent video frames it is determined whether the similarity value of the target object in the adjacent video frames meets a preset similarity threshold according to the first comparison result and the second comparison result. , The target object in the adjacent video frames is taken as the same target object for positioning and tracking.
  • the similarity function is defined as follows:
  • the color feature similarity of the corresponding target objects in each video frame is Sim(b i , b j ); Sim(P i , P j ) is the square of the Euclidean distance between the two points P i and P j .
  • the similarity threshold T Pre-set the similarity threshold T.
  • the similarity Sim(player i ,player j ) of the target object in two adjacent video frames is equal to or greater than T
  • the two adjacent video frames can be regarded as the same target object, and
  • the target objects in two adjacent video frames are identified as the same target object, and the trajectories are merged to realize accurate target object recognition and tracking.
  • multiple target objects in motion can be quickly identified and tracked at the same time, which improves tracking of multiple target objects in motion in video data.
  • the present application also provides a device for tracking multiple target objects in a motion state. Since the embodiment of the device is similar to the above method embodiment, the description is relatively simple. For related details, please refer to the description of the above method embodiment section.
  • the following describes an implementation of a multi-target object tracking device in motion The examples are only illustrative. Please refer to FIG. 2, which is a schematic diagram of a multi-target object tracking device in a motion state provided by an embodiment of this application.
  • the multi-target object tracking device in motion state described in this application includes the following parts:
  • the video frame obtaining unit 201 is configured to obtain video frames contained in the video data collected in the video collecting device.
  • the video acquisition device includes video data acquisition equipment such as cameras, video recorders, and image sensors.
  • the video data is video data contained in an independent shot.
  • an independent lens is video data obtained by a continuous shooting process of the video capture device, and the video data is composed of video frames, and a group of continuous video frames can form a lens.
  • a complete video data may include multiple shots, and the acquisition of the video frames included in the video data collected in the video collection device can be specifically implemented in the following ways:
  • Video frames To obtain the video data collected in the video capture device, before obtaining the video frame contained in one of the shots, it is necessary to first perform shot segmentation on the complete video data based on the global and local features of the video frame to obtain a series of independent Video clips. The similarity between the video clips is detected, and the video clips whose similarity reaches or exceeds the preset similarity threshold and the time interval does not exceed the preset time threshold are taken as a video shot, and then the video shots contained in the video shots are obtained Video frames.
  • the color characteristics of the video frames contained in different shots usually have obvious differences.
  • the color feature extraction algorithm can extract the RGB or HSV color histogram of each video frame in the video data, and then use the window function to calculate the probability distribution of the first half and the second half of the video frame. If the two probabilities are different, the current The center of the window is the lens boundary.
  • the first comparison unit 202 is configured to send the video frame to a preset feature recognition model, determine the feature detection area corresponding to the target object in the video frame, and extract the target object from the detection area
  • the color feature is to compare the color features of the target object in the adjacent video frames to obtain a first comparison result.
  • the feature recognition model may refer to the Faster RCNN deep neural network model.
  • the feature detection area may refer to the detection frame of each target object in the video frame obtained by using the Faster RCNN deep neural network model to perform target object detection on the video frame.
  • RGB red, green, blue
  • HSV Hue, Saturation, Value, hue, saturation, brightness
  • the second comparison unit 203 is configured to determine the position information of the identification part of the target object in the adjacent video frame in the target coordinate system, and place the identification part in the adjacent video frame in the target coordinate system. Compare the location information in to obtain the second comparison result.
  • the target coordinate system may refer to the world coordinate system, which may refer to the absolute coordinate system of the video screen, and the coordinates of the points corresponding to the identification parts of all target objects in the video screen can be
  • the world coordinate system is used to determine the specific location of each target object.
  • the world coordinate system may refer to a spatial rectangular coordinate system constructed with the center of the detection area as the origin of the spatial coordinate system.
  • point P can refer to the position of the point corresponding to the neck of the target object
  • point Q1 can refer to the position of the point corresponding to the video capture device in the previous video frame, or it can refer to the point corresponding to the video capture device The position of the point in the previous shot
  • Q2 point can refer to the position of the point corresponding to the video capture device in the next video frame relative to the previous video, or the point corresponding to the video capture device The position in the next lens relative to the previous lens.
  • the determination of the position information of the identification part of the target object in the adjacent video frame in the target coordinate system may be specifically implemented in the following manner:
  • the visual mileage calculation method can be used to predict the pose change of the video acquisition device, and the corresponding adjacent video frames can be obtained through prediction.
  • the position and posture change information of the video capture device is further obtained, and the posture change information of the video capture device corresponding to the adjacent video frames is obtained. According to the posture change information, the position information of the video capture devices corresponding to the adjacent video frames can be determined.
  • the position information of the video acquisition device in the previous video frame in the adjacent video frames may be recorded as the first position, and the position information of the video acquisition device in the next video frame in the adjacent video frames Recorded as the second position.
  • the target object can be obtained by calculation using the triangulation method shown in FIG.
  • the position information in the spatial rectangular coordinate system constructed with the video capture device as the spatial coordinate origin can be further obtained by coordinate transformation to obtain the position information of the target object in the target coordinate system (ie, the world coordinate system).
  • the posture changes include changes in motion trajectories and activity postures.
  • the identification part may select the neck part of the target object.
  • the position information of the identification part of the target object in the target coordinate system is the position information of the neck part in the spatial rectangular coordinate system constructed with the center of the area to be detected as the spatial coordinate origin.
  • a bone detection algorithm can be used to obtain the point P corresponding to the neck of each target object.
  • the judging unit 204 is configured to judge whether the target objects in the adjacent video frames are the same target object according to the first comparison result and the second comparison result, and if so, compare the adjacent video frames The target object mentioned in is tracked as the same target object.
  • the target object in the adjacent video frames it is determined whether the similarity value of the target object in the adjacent video frames meets a preset similarity threshold according to the first comparison result and the second comparison result. , The target object in the adjacent video frames is taken as the same target object for positioning and tracking.
  • the similarity function is defined as follows:
  • the color feature similarity of the corresponding target objects in each video frame is Sim(b i , b j ); Sim(P i , P j ) is the square of the Euclidean distance between the two points P i and P j .
  • the multi-target object tracking device in motion state described in this application can simultaneously quickly identify and track multi-target objects in motion state, which improves the tracking of multi-target objects in motion state in video data. The accuracy of, thereby enhancing the user experience.
  • the present application also provides an electronic device and a storage device. Since the embodiment of the electronic device is similar to the above method embodiment, the description is relatively simple. For related details, please refer to the description of the above method embodiment.
  • the following describes an embodiment of an electronic device and a storage device. The examples are only illustrative. Please refer to FIG. 4, which is a schematic diagram of an electronic device provided by an embodiment of this application.
  • the present application also provides an electronic device, including: a processor 401 and a memory 402; wherein the memory 402 is used to store a program for the tracking method for multiple targets in a motion state, and the device is powered on and runs through the processor After the procedure for the multi-target tracking method in motion, the following steps are performed:
  • the position information of the marking part in the target coordinate system is compared with the position information of the marking part in the adjacent video frames in the target coordinate system to obtain a second comparison result; according to the first comparison result and
  • the second comparison result determines whether the target object in the adjacent video frame is the same target object, and if so, the target object in the adjacent video frame is tracked as the same target object.
  • This application also provides a storage device that stores a program for the tracking method for multiple targets in a motion state, and the program is run by the processor to perform the following steps:
  • the position information of the marking part in the target coordinate system is compared with the position information of the marking part in the adjacent video frames in the target coordinate system to obtain a second comparison result; according to the first comparison result and
  • the second comparison result determines whether the target object in the adjacent video frame is the same target object, and if so, the target object in the adjacent video frame is tracked as the same target object.
  • the processor or the processor module may be an integrated circuit chip with signal processing capability.
  • the processor can be a general-purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (Field Programmable Gate Array, FPGA) or other Programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.
  • DSP digital signal processor
  • ASIC application specific integrated circuit
  • FPGA field programmable gate array
  • Programmable logic devices discrete gate or transistor logic devices, discrete hardware components.
  • the methods, steps, and logical block diagrams disclosed in the embodiments of the present application can be implemented or executed.
  • the general-purpose processor may be a microprocessor or the processor may also be any conventional processor or the like.
  • the steps of the method disclosed in the embodiments of the present application may be directly embodied as being executed and completed by a hardware decoding processor, or executed and completed by a combination of hardware and software modules in the decoding processor.
  • the software module can be located in a mature storage medium in the field such as random access memory, flash memory, read-only memory, programmable read-only memory, or electrically erasable programmable memory, registers.
  • the processor reads the information in the storage medium and completes the steps of the above method in combination with its hardware.
  • the storage medium may be a memory, for example, may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memory.
  • the non-volatile memory can be a read-only memory (Read-Only Memory, ROM for short), a programmable read-only memory (Programmable ROM, PROM for short), and an erasable programmable read-only memory (Erasable PROM, EPROM for short). , Electrically Erasable Programmable Read-Only Memory (Electrically EPROM, EEPROM for short) or flash memory.
  • Read-Only Memory Read-Only Memory
  • PROM programmable read-only memory
  • Erasable PROM Erasable PROM, EPROM for short
  • Electrically Erasable Programmable Read-Only Memory Electrically Erasable Programmable Read-Only Memory (Electrically EPROM, EEPROM for short) or flash memory.
  • the volatile memory may be a random access memory (Random Access Memory, RAM for short), which is used as an external cache.
  • RAM Random Access Memory
  • many forms of RAM are available, such as static random access memory (Static RAM, SRAM for short), dynamic random access memory (Dynamic RAM, DRAM for short), and synchronous dynamic random access memory (Synchronous RAM).
  • DRAM static random access memory
  • DRAM dynamic random access memory
  • DRAM dynamic random access memory
  • Synchronous RAM Synchronous Dynamic Random Access Memory
  • DRAM Double Data Rate Synchronous Dynamic Random Access Memory
  • Enhanced SDRAM for ESDRAM
  • Synch link DRAM, SLDRAM for short Synchronously Connected Dynamic Random Access Memory
  • DRRAM Direct Ram bus RAM
  • the storage media described in the embodiments of the present application are intended to include, but are not limited to, these and any other suitable types of memory.
  • the functions described in this application can be implemented by a combination of hardware and software.
  • the corresponding function can be stored in a computer-readable medium or transmitted as one or more instructions or codes on the computer-readable medium.
  • the computer-readable medium includes a computer storage medium and a communication medium, where the communication medium includes any medium that facilitates the transfer of a computer program from one place to another.
  • the storage medium may be any available medium that can be accessed by a general-purpose or special-purpose computer.

Landscapes

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

Abstract

本申请实施例公开了一种针对运动状态下的多目标对象追踪方法和装置，其中，所述方法包括：从视频采集装置采集的视频帧中确定目标对象的特征检测区域，从所述检测区域中提取所述目标对象的颜色特征进行比对，获得第一比对结果；将相邻所述视频帧中所述目标对象的标识部位在目标坐标系中的位置信息进行比对，获得第二比对结果；根据所述第一比对结果和所述第二比对结果，判断相邻所述视频帧中的所述目标对象是否为同一目标对象，进而实现精确的定位和追踪。采用本申请所述的针对运动状态下的多目标对象追踪方法，能够同时对多目标对象进行快速识别和追踪，提高了针对视频数据中目标对象进行识别和追踪的精确度。

Description

一种针对运动状态下的多目标对象追踪方法和装置
本申请要求于2018年06月17日提交中国专利局，申请号为CN201910522911.3，申请名称为“一种针对运动状态下的多目标对象追踪方法和装置”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请实施例涉及人工智能技术领域，具体涉及一种针对运动状态下的多目标对象追踪方法和装置。
背景技术
随着计算机视觉技术的快速发展，现有视频采集设备的功能越来越强大，用户可以通过视频采集设备实现对视频数据中特定目标对象的追踪拍摄。计算机视觉技术是一种研究如何使机器“看”的技术，可以用摄影机和计算机设备代替人眼对目标对象进行实时识别、定位、跟踪和测量等机器视觉处理技术。通过计算机设备对图像进行分析处理，使摄影机获取的数据更适合人眼观察或传送给仪器检测的图像信息。例如，在篮球比赛中，通常需要利用摄像机对场上的多名球员同时进行追踪拍摄，使用户可以根据需求随时切换至球员相应的追踪拍摄角度或者获得该球员在场上的运动轨迹数据。因此，如何在视频采集设备和目标对象都处于运动状态的情况下实现对目标对象的快速、精确定位和追踪，成为当前亟待解决的技术问题。
为了解决上述技术问题，现有技术中通常采用的技术手段是基于2D图像识别技术对视频帧中的目标对象的位置相似度进行判断，确定相邻视频帧中的目标对象是否为同一目标对象，进而实现对目标对象的定位追踪以及获得所述目标对象的运动轨迹。然而，在实际应用场景中，除 了目标对象处于运动状态之外，往往还存在视频采集设备本身的位姿变化，导致现有技术对目标对象的实际追踪拍摄效果较差，容易出现识别错误，无法满足当前用户的需求。
发明内容
为此，本申请实施例提供一种针对运动状态下的多目标对象追踪方法，以解决现有技术中存在针对视频中多目标对象的识别和追踪效率较低、精确度较差的问题。
为了实现上述目的，本申请实施例提供如下技术方案：
根据本申请实施例提供一种针对运动状态下的多目标对象追踪方法，包括：获得视频采集装置中采集的视频数据所包含的视频帧；将所述视频帧发送至预设的特征识别模型中，确定所述视频帧中对应目标对象的特征检测区域，从所述检测区域中提取所述目标对象的颜色特征，将相邻所述视频帧中所述目标对象的所述颜色特征进行比对，获得第一比对结果；确定相邻所述视频帧中所述目标对象的标识部位在目标坐标系中的位置信息，将相邻所述视频帧中的所述标识部位在目标坐标系中的位置信息进行比对，获得第二比对结果；根据所述第一比对结果和所述第二比对结果，判断相邻所述视频帧中的所述目标对象是否为同一目标对象，若是，则将相邻所述视频帧中所述目标对象作为同一目标对象进行追踪。
进一步的，所述确定相邻所述视频帧中所述目标对象的标识部位在目标坐标系中的位置信息，具体包括：通过预测相邻所述视频帧分别对应的所述视频采集装置的位姿变化情况，获得相邻所述视频帧中分别对应的所述视频采集装置的位姿变化信息；根据所述位姿变化信息和相邻所述视频帧中前一视频帧对应的所述视频采集装置的位置信息，确定相邻所述视频帧中后一视频帧对应的所述视频采集装置的位置信息；根据相邻所述视频帧中分别对应的所述视频采集装置的位置信息以及所述目标对象的标识部位，利用三角测量法获得所述目标对象的标识部位在以 所述视频采集装置为空间坐标原点构建的空间直角坐标系中的位置信息；通过坐标变换获得所述目标对象的标识部位在目标坐标系中的位置信息。
进一步的，所述的针对运动状态下的多目标对象追踪方法，还包括：确定所述视频帧中所述目标对象的实际运动区域；将所述视频帧中所述目标对象的实际运动区域作为待检测区域，对所述待检测区域之外的所述特征检测区域进行滤除，获得所述待检测区域之内的所述特征检测区域。
进一步的，所述标识部位为所述目标对象的颈部部位；相应的，所述目标对象的标识部位在目标坐标系中的位置信息为所述目标对象的颈部部位位于以所述待检测区域中心为空间坐标原点构建的空间直角坐标系中的位置信息。
进一步的，获得所述视频采集装置中采集的所述视频数据，对所述视频数据进行分割处理，获得所述视频数据所包含的视频片段；检测所述视频片段之间的特征相似度，将所述特征相似度达到或超过预设相似度阈值并且时间间隔不超过预设时间阈值的视频片段作为一个视频镜头；获取所述视频镜头中所包含的视频帧。
相应的，本申请实施例还提供一种针对运动状态下的多目标对象追踪装置，包括：视频帧获得单元，用于获得视频采集装置中采集的视频数据所包含的视频帧；第一比对单元，用于将所述视频帧发送至预设的特征识别模型中，确定所述视频帧中对应目标对象的特征检测区域，从所述检测区域中提取所述目标对象的颜色特征，将相邻所述视频帧中所述目标对象的所述颜色特征进行比对，获得第一比对结果；第二比对单元，用于确定相邻所述视频帧中所述目标对象的标识部位在目标坐标系中的位置信息，将相邻所述视频帧中的所述标识部位在目标坐标系中的位置信息进行比对，获得第二比对结果；判断单元，用于根据所述第一比对结果和所述第二比对结果，判断相邻所述视频帧中的所述目标对象是否为同一目标对象，若是，则将相邻所述视频帧中所述目标对象作为 同一目标对象进行追踪。
进一步的，所述确定相邻所述视频帧中所述目标对象的标识部位在目标坐标系中的位置信息，具体包括：通过预测相邻所述视频帧分别对应的所述视频采集装置的位姿变化情况，获得相邻所述视频帧中分别对应的所述视频采集装置的位姿变化信息；根据所述位姿变化信息和相邻所述视频帧中前一视频帧对应的所述视频采集装置的位置信息，确定相邻所述视频帧中后一视频帧对应的所述视频采集装置的位置信息；根据相邻所述视频帧中分别对应的所述视频采集装置的位置信息以及所述目标对象的标识部位，利用三角测量法获得所述目标对象的标识部位在以所述视频采集装置为空间坐标原点构建的空间直角坐标系中的位置信息；通过坐标变换获得所述目标对象的标识部位在目标坐标系中的位置信息。
进一步的，所述的针对运动状态下的多目标对象追踪装置，还包括：运动区域确定单元，用于确定所述视频帧中所述目标对象的实际运动区域；滤除单元，用于将所述视频帧中所述目标对象的实际运动区域作为待检测区域，对所述待检测区域之外的所述特征检测区域进行滤除，获得所述待检测区域之内的所述特征检测区域。
进一步的，所述标识部位为所述目标对象的颈部部位；相应的，所述目标对象的标识部位在目标坐标系中的位置信息为所述目标对象的颈部部位位于以所述待检测区域中心为空间坐标原点构建的空间直角坐标系中的位置信息。
进一步的，所述获得视频采集装置中采集的视频数据所包含的视频帧，具体包括：获得所述视频采集装置中采集的所述视频数据，对所述视频数据进行分割处理，获得所述视频数据所包含的视频片段；检测所述视频片段之间的特征相似度，将所述特征相似度达到或超过预设相似度阈值并且时间间隔不超过预设时间阈值的视频片段作为一个视频镜头；获取所述视频镜头中所包含的视频帧。
相应的，本申请还提供一种电子设备，包括：处理器和存储器；其 中，所述存储器用于存储针对运动状态下的多目标对象追踪方法的程序，该设备通电并通过所述处理器运行该针对运动状态下的多目标对象追踪方法的程序后，执行下述步骤：
获得视频采集装置中采集的视频数据所包含的视频帧；将所述视频帧发送至预设的特征识别模型中，确定所述视频帧中对应目标对象的特征检测区域，从所述检测区域中提取所述目标对象的颜色特征，将相邻所述视频帧中所述目标对象的所述颜色特征进行比对，获得第一比对结果；确定相邻所述视频帧中所述目标对象的标识部位在目标坐标系中的位置信息，将相邻所述视频帧中的所述标识部位在目标坐标系中的位置信息进行比对，获得第二比对结果；根据第一比对结果和所述第二比对结果，判断相邻所述视频帧中的所述目标对象是否为同一目标对象，若是，则将相邻所述视频帧中所述目标对象作为同一目标对象进行追踪。
相应的，本申请还提供一种存储设备，存储有针对运动状态下的多目标对象追踪方法的程序，该程序被处理器运行，执行下述步骤：
获得视频采集装置中采集的视频数据所包含的视频帧；将所述视频帧发送至预设的特征识别模型中，确定所述视频帧中对应目标对象的特征检测区域，从所述检测区域中提取所述目标对象的颜色特征，将相邻所述视频帧中所述目标对象的所述颜色特征进行比对，获得第一比对结果；确定相邻所述视频帧中所述目标对象的标识部位在目标坐标系中的位置信息，将相邻所述视频帧中的所述标识部位在目标坐标系中的位置信息进行比对，获得第二比对结果；根据第一比对结果和所述第二比对结果，判断相邻所述视频帧中的所述目标对象是否为同一目标对象，若是，则将相邻所述视频帧中所述目标对象作为同一目标对象进行追踪。
采用本申请所述的针对运动状态下的多目标对象追踪方法，能够同时对处于运动状态下的多目标对象进行快速的识别和追踪，提高了针对视频数据中处于运动状态的多目标对象进行识别和追踪的精确度，从而提升了用户的使用体验。
附图说明
为了更清楚地说明本申请的实施方式或现有技术中的技术方案，下面将对实施方式或现有技术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示例性的，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，还可以根据提供的附图引伸获得其它的实施附图。
本说明书所绘示的结构、比例、大小等，均仅用以配合说明书所揭示的内容，以供熟悉此技术的人士了解与阅读，并非用以限定本申请可实施的限定条件，故不具技术上的实质意义，任何结构的修饰、比例关系的改变或大小的调整，在不影响本申请所能产生的功效及所能达成的目的下，均应仍落在本申请所揭示的技术内容得能涵盖的范围内。
图1为本申请实施例提供的一种针对运动状态下的多目标对象追踪方法的流程图；
图2为本申请实施例提供的一种针对运动状态下的多目标对象追踪装置的示意图；
图3为本申请实施例提供的一种利用三角测量法定位目标对象的示意图；
图4为本申请实施例提供的一种电子设备的示意图。
具体实施方式
以下由特定的具体实施例说明本申请的实施方式，熟悉此技术的人士可由本说明书所揭露的内容轻易地了解本申请的其他优点及功效，显然，所描述的实施例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做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。
下面基于本申请所述的针对运动状态下的多目标对象追踪方法，对其实施例进行详细描述。如图1所示，其为本申请实施例提供的一种针对运动状态下的多目标对象追踪方法的流程图，具体实现过程包括以下 步骤：
步骤S101：获得视频采集装置中采集的视频数据所包含的视频帧。
在本申请实施例中，所述的视频采集装置包括摄像机、录像机以及图像传感器等视频数据采集设备。所述的视频数据为一个独立镜头内所包含的视频数据。其中，一个独立镜头是视频采集装置的一个连续拍摄过程获得的视频数据，视频数据由视频帧画面组成，一组连续的视频帧可以构成一个镜头。
在一个完整的视频数据中可能包含多个镜头，所述的获得视频采集装置中采集的视频数据所包含的视频帧，具体可以通过如下方式实现：
获得所述视频采集装置中采集的所述视频数据，在获取其中一个镜头所包含的视频帧之前需要首先基于视频帧的全局特征和局部特征对完整的视频数据进行镜头分割，得到一系列独立的视频片段。检测所述视频片段之间的相似度，将所述相似度达到或超过预设相似度阈值并且时间间隔不超过预设时间阈值的视频片段作为一个视频镜头，进而获取所述视频镜头中所包含的视频帧。
在具体实施过程中，不同镜头所包含的视频帧的颜色特征通常存在明显差异，当相邻两个视频帧之间的颜色特征发生变化时，则可以认为在此处发生了镜头的切换，利用颜色特征提取算法可以提取视频数据中每一视频帧的RGB或HSV颜色直方图，然后利用窗口函数计算视频帧画面中前半部分和后半部分的概率分布，若两个概率不同则认为此时的窗口中心为镜头分界。
所述的基于视频帧的全局特征和局部特征对完整的视频数据进行镜头分割，具体可以通过如下过程实现：
全局特征分析：基于相邻视频帧颜色特征计算视频数据的相邻视频帧之间的第一相似度，将所述第一相似度与第一相似度阈值进行比较，若所述第一相似度小于所述第一相似度阈值，则将该视频帧作为一个独立镜头的候选视频帧。
局部特征分析：分别计算所述候选视频帧与其前一视频帧中关键点 的描述子到每一个视觉词的距离值，将描述子与所述视觉词的距离值最小的视觉词相对应，基于描述子和对应的视觉词，分别构建所述候选视频帧和其前一帧的视觉词直方图，计算视频帧的视觉词直方图之间的第二相似度。
镜头分割步骤：对所述第二相似度进行判断，若所述第二相似度大于或者等于第二相似度阈值，则将所述候选视频帧和其前一帧归并为同一个镜头，若所述第二相似度小于所述第二相似度阈值，则将所述候选视频帧确定为新镜头的起始视频帧。
步骤S102：将所述视频帧发送至预设的特征识别模型中，确定所述视频帧中对应目标对象的特征检测区域，从所述特征检测区域中提取所述目标对象的颜色特征，将相邻所述视频帧中所述目标对象的所述颜色特征进行比对，获得第一比对结果。
在上述步骤S101中获得视频采集装置采集的视频数据中所包含的视频帧后，为本步骤将相邻视频帧中目标对象的颜色特征进行比对做了数据准备工作。在步骤S102中，可以从所述视频帧中提取目标对象的颜色特征，进一步将相邻所述视频帧中所述目标对象的所述颜色特征进行比对，获得第一比对结果。
在本申请实施例中，所述的特征识别模型可以是指预先经过迭代训练获得的Faster RCNN深度神经网络模型。所述的特征检测区域可以是指针对视频帧使用Faster RCNN深度神经网络模型进行目标对象检测过程中，得到的视频画面中的每个目标对象对应的检测框。
具体的，考虑到相邻视频帧中对应每个目标对象的待检测区域中的每个像素位置的RGB(红、绿、蓝)颜色或HSV(Hue Saturation Value，色调、饱和度、明度)颜色的通常相同或相似，因此，可以从所述待检测区域中提取所述目标对象的颜色特征，将相邻所述视频帧中所述目标对象的所述颜色特征进行比对，获得第一比对结果，即：相邻所述视频帧中所述目标对象的所述颜色特征之间的相似度。
考虑到在实际实施过程中，确定所述视频帧中对应目标对象的特征 检测区域时，最终确定的检测结果中可能存在对应非目标对象生成的检测区域(即非目标对象对应的监测框)，此时，需要预先对上述检测结果进行过滤，只保留对应目标对象的特征检测区域(即目标对象对应的监测框)，具体实现方式如下：
确定所述视频帧中所述目标对象的实际运动区域，将所述视频帧中所述目标对象的实际运动区域作为待检测区域，对所述待检测区域之外的所述特征检测区域进行滤除，获得所述待检测区域之内的所述特征检测区域。其中，所述的实际运动区域，即：所述目标对象的活动区域。
以篮球比赛为例对上述实现方式进行说明：在篮球比赛过程中，首先需要针对每一视频帧中包含的球员使用特征识别模型进行球员检测，获得视频画面中每个球员(即目标对象)分别对应的检测框，并记录唯一标识球员身份的ID。此时，可能将球场之外的观众(即非目标对象)也会产生相应的检测框，然而，观众并非需要定位和追踪的目标对象，因此需要将对应观众的检测框进行滤除，仅保留球场范围内的检测框。具体的，可以利用球场地板的颜色特征和观众席之间颜色特征的差异，通过阈值过滤法进行区分过滤，得到仅包含球场的图像，进一步对球场的图像进行腐蚀、膨胀等一系列处理操作，得到球场的外轮廓(所述外轮廓围成的区域即所述目标对象的实际运动区域)，将球场的外轮廓之外的检测框进行滤除，仅保外轮廓围成的区域之内(即：留球场之内)的检测框。
步骤S103：确定相邻所述视频帧中所述目标对象的标识部位在目标坐标系中的位置信息，将相邻所述视频帧中的所述标识部位在目标坐标系中的位置信息进行比对，获得第二比对结果。
上述步骤S102中获得第一比对结果后，本步骤可以进一步确定相邻所述视频帧中所述目标对象的标识部位在目标坐标系中的位置信息，将相邻所述视频帧中的所述标识部位在目标坐标系中的位置信息进行比对，获得第二比对结果。
在本申请实施例中，所述的目标坐标系可以是指世界坐标系，世界 坐标系可以是指视频画面的绝对坐标系，在视频画面中所有目标对象的标识部位对应的点的坐标都可以以该世界坐标系来确定各个目标对象所处的具体位置。其中，所述的世界坐标系可以是指以检测区域中心为空间坐标系原点构建的空间直角坐标系。
如图3所示，为本申请实施例提供的一种利用三角测量法定位目标对象的示意图。其中，P点可以是指目标对象的颈部部位对应的点的位置；Q1点可以是指视频采集装置对应的点在前一视频帧中所处的位置，也可以是指视频采集装置对应的点在前一镜头中所处的位置；Q2点可以是指视频采集装置对应的点在相对于前一视频的后一视频帧中所处的位置，也可以是指视频采集装置对应的点在相对于前一镜头的后一镜头中所处的位置。
所述的确定相邻所述视频帧中所述目标对象的标识部位在目标坐标系中的位置信息，具体可以通过如下方式实现：
首先，对上述完整视频数据中的每一个镜头，可以利用视觉里程计算法(特征点法)对视频采集装置的位姿变化进行预测，通过预测可以获得相邻所述视频帧分别对应的所述视频采集装置的位姿变化情况，进而获得相邻所述视频帧中分别对应的所述视频采集装置的位姿变化信息。根据位姿变化信息，可以确定相邻所述视频帧中分别对应的所述视频采集装置的位置信息。
在此，可以将相邻所述视频帧中的前一视频帧中视频采集装置的位置信息记为第一位置，将相邻所述视频帧中的后一视频帧中视频采集装置的位置信息记为第二位置。
根据相邻所述视频帧中视频采集装置分别对应的第一位置、第二位置以及所述标识部位对应的点的位置，利用如图3所示的三角测量法进行计算可以获得所述目标对象在以所述视频采集装置为空间坐标原点构建的空间直角坐标系中的位置信息，进一步通过坐标变换即可以获得所述目标对象在目标坐标系(即：世界坐标系)中的位置信息。其中，所述位姿变化包括运动轨迹和活动姿态的变化情况等。
需要说明的是，为了便于对目标对象进行精确的定位和追踪，所述标识部位可以选择所述目标对象的颈部部位。所述目标对象的标识部位在目标坐标系中的位置信息为所述颈部部位位于以所述待检测区域中心为空间坐标原点构建的空间直角坐标系中的位置信息。具体的，在特征检测区域中，可以使用骨骼检测算法得到每个目标对象的颈部部位对应的点P。
步骤S104：根据第一比对结果和所述第二比对结果，判断相邻所述视频帧中的所述目标对象是否为同一目标对象，若是，则将相邻所述视频帧中所述目标对象作为同一目标对象进行追踪。
上述步骤S102和步骤S103中分别获得第一比对结果和第二比对结果后，本步骤可以根据第一比对结果和所述第二比对结果，判断相邻所述视频帧中的所述目标对象是否为同一目标对象，进而实现对目标对象实时的定位和追踪。
在本申请实施例中，所述根据第一比对结果和所述第二比对结果，判断相邻所述视频帧中的所述目标对象的相似度值是否满足预设相似度阈值，若是，则将相邻所述视频帧中的所述目标对象作为同一目标对象进行定位和追踪。
具体的，根据相邻两个视频帧中目标对象所对应的颜色特征和位置信息之间的相似性，采用两两比对的方式，利用相似性函数可以进行计算，定义相似性函数如下：
Sim(player i,player j)＝-(Sim(b i,b j)+Sim(P i,P j))；
其中，Sim(player i,player j)为相邻两个视频帧中目标对象的相似度；记录相邻两个视频帧中每个目标对象为player i＝(b i，P i)；Sim(b i,b j)＝|f(b i)-f(b j)|，其中函数f为外观特征提取函数，使用方向梯度直方图(Histogram of Oriented Gradient，HOG)的方式可以获得相邻两个视频帧中对应目标对象的颜色特征相似度Sim(b i,b j)；Sim(P i,P j)为两点P i、P j的欧氏距离的平方。
预先设定相似度阈值T，当相邻两个视频帧中目标对象的相似度 Sim(player i,player j)等于或大于T时，可以认定相邻两个视频帧为同一目标对象，并将相邻两两视频帧中的目标对象认定为同一目标对象，进行轨迹合并，从而实现精确的目标对象识别和追踪。
采用本申请所述的针对运动状态下的多目标对象追踪方法，能够同时对处于运动状态下的多目标对象进行快速的识别和追踪，提高了针对视频数据中处于运动状态的多目标对象进行追踪的精确度，从而提升了用户的使用体验。
与上述提供的一种针对运动状态下的多目标对象追踪方法相对应，本申请还提供一种针对运动状态下的多目标对象追踪装置。由于该装置的实施例相似于上述方法实施例，所以描述的比较简单，相关之处请参见上述方法实施例部分的说明即可，下面描述一种针对运动状态下的多目标对象追踪装置的实施例仅是示意性的。请参考图2所示，其为本申请实施例提供的一种针对运动状态下的多目标对象追踪装置的示意图。
本申请所述的一种针对运动状态下的多目标对象追踪装置包括如下部分：
视频帧获得单元201，用于获得视频采集装置中采集的视频数据所包含的视频帧。
在本申请实施例中，所述的视频采集装置包括摄像机、录像机以及图像传感器等视频数据采集设备。所述的视频数据为一个独立镜头内所包含的视频数据。其中，一个独立镜头是视频采集装置的一个连续拍摄过程获得的视频数据，视频数据由视频帧画面组成，一组连续的视频帧可以构成一个镜头。
在一个完整的视频数据中可能包含多个镜头，所述的获得视频采集装置中采集的视频数据所包含的视频帧，具体可以通过如下方式实现：
获得所述视频采集装置中采集的所述视频数据，在获取其中一个镜头所包含的视频帧之前需要首先基于视频帧的全局特征和局部特征对完整的视频数据进行镜头分割，得到一系列独立的视频片段。检测所述视频片段之间的相似度，将所述相似度达到或超过预设相似度阈值并且时 间间隔不超过预设时间阈值的视频片段作为一个视频镜头，进而获取所述视频镜头中所包含的视频帧。
在具体实施过程中，不同镜头所包含的视频帧的颜色特征通常存在明显差异，当相邻两个视频帧之间的颜色特征发生变化时，则可以认为在此处发生了镜头的切换，利用颜色特征提取算法可以提取视频数据中每一视频帧的RGB或HSV颜色直方图，然后利用窗口函数计算视频帧画面中前半部分和后半部分的概率分布，若两个概率不同则认为此时的窗口中心为镜头分界。
第一比对单元202，用于将所述视频帧发送至预设的特征识别模型中，确定所述视频帧中对应目标对象的特征检测区域，从所述检测区域中提取所述目标对象的颜色特征，将相邻所述视频帧中所述目标对象的所述颜色特征进行比对，获得第一比对结果。
在本申请实施例中，所述的特征识别模型可以是指Faster RCNN深度神经网络模型。所述的特征检测区域可以是指针对视频帧使用Faster RCNN深度神经网络模型进行目标对象检测，得到的视频画面中的每个目标对象的检测框。
具体的，考虑到相邻视频帧中对应每个目标对象的待检测区域中的每个像素位置的RGB(红、绿、蓝)颜色或HSV(Hue，Saturation，Value，色调、饱和度、明度)颜色的通常相同或相似，因此，可以从所述待检测区域中提取所述目标对象的颜色特征，将相邻所述视频帧中所述目标对象的所述颜色特征进行比对，获得第一比对结果。
考虑到在实际实施过程中，确定所述视频帧中对应目标对象的特征检测区域时，最终确定的检测结果中可能存在对应非目标对象生成的检测区域，此时，需要对上述检测结果进行过滤，只保留对应目标对象的特征检测区域，具体实现方式如下：
确定所述视频帧中所述目标对象的实际运动区域，将所述视频帧中所述目标对象的实际运动区域作为待检测区域，对所述待检测区域之外的所述特征检测区域进行滤除，获得所述待检测区域之内的所述特征检 测区域。
第二比对单元203，用于确定相邻所述视频帧中所述目标对象的标识部位在目标坐标系中的位置信息，将相邻所述视频帧中的所述标识部位在目标坐标系中的位置信息进行比对，获得第二比对结果。
在本申请实施例中，所述的目标坐标系可以是指世界坐标系，世界坐标系可以是指视频画面的绝对坐标系，在视频画面中所有目标对象的标识部位对应的点的坐标都可以以该世界坐标系来确定各个目标对象所处的具体位置。其中，所述的世界坐标系可以是指以检测区域中心为空间坐标系原点构建的空间直角坐标系。
如图3所示，为本申请实施例提供的一种利用三角测量法定位目标对象的示意图。其中，P点可以是指目标对象的颈部部位对应的点的位置；Q1点可以是指视频采集装置对应的点在前一视频帧中所处的位置，也可以是指视频采集装置对应的点在前一镜头中所处的位置；Q2点可以是指视频采集装置对应的点在相对于前一视频的后一视频帧中所处的位置，也可以是指视频采集装置对应的点在相对于前一镜头的后一镜头中所处的位置。
所述的确定相邻所述视频帧中所述目标对象的标识部位在目标坐标系中的位置信息，具体可以通过如下方式实现：
首先，对上述完整视频数据中的每一个镜头，可以利用视觉里程计算法(特征点法)对视频采集装置的位姿变化进行预测，通过预测可以获得相邻所述视频帧分别对应的所述视频采集装置的位姿变化情况，进而获得相邻所述视频帧中分别对应的所述视频采集装置的位姿变化信息。根据位姿变化信息，可以确定相邻所述视频帧中分别对应的所述视频采集装置的位置信息。
在此，可以将相邻所述视频帧中的前一视频帧中视频采集装置的位置信息记为第一位置，将相邻所述视频帧中的后一视频帧中视频采集装置的位置信息记为第二位置。
根据相邻所述视频帧中视频采集装置分别对应的第一位置、第二位 置以及所述标识部位对应的点的位置，利用如图3所示的三角测量法进行计算可以获得所述目标对象在以所述视频采集装置为空间坐标原点构建的空间直角坐标系中的位置信息，进一步通过坐标变换即可以获得所述目标对象在目标坐标系(即：世界坐标系)中的位置信息。其中，所述位姿变化包括运动轨迹和活动姿态的变化情况等。
需要说明的是，为了便于对目标对象进行精确的定位和追踪，所述标识部位可以选择所述目标对象的颈部部位。所述目标对象的标识部位在目标坐标系中的位置信息为所述颈部部位位于以所述待检测区域中心为空间坐标原点构建的空间直角坐标系中的位置信息。具体的，在特征检测区域中，可以使用骨骼检测算法得到每个目标对象的颈部部位对应的点P。
判断单元204，用于根据第一比对结果和所述第二比对结果，判断相邻所述视频帧中的所述目标对象是否为同一目标对象，若是，则将相邻所述视频帧中所述目标对象作为同一目标对象进行追踪。
在本申请实施例中，所述根据第一比对结果和所述第二比对结果，判断相邻所述视频帧中的所述目标对象的相似度值是否满足预设相似度阈值，若是，则将相邻所述视频帧中的所述目标对象作为同一目标对象进行定位和追踪。
具体的，根据相邻两个视频帧中目标对象所对应的颜色特征和位置信息之间的相似性，采用两两比对的方式，利用相似性函数可以进行计算，定义相似性函数如下：
Sim(player i,player j)＝-(Sim(b i,b j)+Sim(P i,P j))；
其中，Sim(player i,player j)为相邻两个视频帧中目标对象的相似度；记录相邻两个视频帧中每个目标对象为player i＝(b i，P i)；Sim(b i,b j)＝|f(b i)-f(b j)|，其中函数f为外观特征提取函数，使用方向梯度直方图(Histogram of Oriented Gradient，HOG)的方式可以获得相邻两个视频帧中对应目标对象的颜色特征相似度Sim(b i,b j)；Sim(P i,P j)为两点P i、P j的欧氏距离的平方。
预先设定相似度阈值T，当相邻两个视频帧中目标对象的相似度Sim(player i,player j)等于或大于T时，可以认定相邻两个视频帧为同一目标对象，并进行轨迹合并。
采用本申请所述的针对运动状态下的多目标对象追踪装置，能够同时对处于运动状态下的多目标对象进行快速的识别和追踪，提高了针对视频数据中处于运动状态的多目标对象进行追踪的精确度，从而提升了用户的使用体验。
与上述提供的一种针对运动状态下的多目标对象追踪方法相对应，本申请还提供一种电子设备和存储设备。由于该电子设备的实施例相似于上述方法实施例，所以描述的比较简单，相关之处请参见上述方法实施例部分的说明即可，下面描述一种电子设备的实施例和一种存储设备的实施例仅是示意性的。请参考图4所示，其为本申请实施例提供的一种电子设备的示意图。
本申请还提供一种电子设备，包括：处理器401和存储器402；其中，所述存储器402用于存储针对运动状态下的多目标对象追踪方法的程序，该设备通电并通过所述处理器运行该针对运动状态下的多目标对象追踪方法的程序后，执行下述步骤：
获得视频采集装置中采集的视频数据所包含的视频帧；将所述视频帧发送至预设的特征识别模型中，确定所述视频帧中对应目标对象的特征检测区域，从所述检测区域中提取所述目标对象的颜色特征，将相邻所述视频帧中所述目标对象的所述颜色特征进行比对，获得第一比对结果；确定相邻所述视频帧中所述目标对象的标识部位在目标坐标系中的位置信息，将相邻所述视频帧中的所述标识部位在目标坐标系中的位置信息进行比对，获得第二比对结果；根据第一比对结果和所述第二比对结果，判断相邻所述视频帧中的所述目标对象是否为同一目标对象，若是，则将相邻所述视频帧中所述目标对象作为同一目标对象进行追踪。
本申请还提供一种存储设备，存储有针对运动状态下的多目标对象追踪方法的程序，该程序被处理器运行，执行下述步骤：
获得视频采集装置中采集的视频数据所包含的视频帧；将所述视频帧发送至预设的特征识别模型中，确定所述视频帧中对应目标对象的特征检测区域，从所述检测区域中提取所述目标对象的颜色特征，将相邻所述视频帧中所述目标对象的所述颜色特征进行比对，获得第一比对结果；确定相邻所述视频帧中所述目标对象的标识部位在目标坐标系中的位置信息，将相邻所述视频帧中的所述标识部位在目标坐标系中的位置信息进行比对，获得第二比对结果；根据第一比对结果和所述第二比对结果，判断相邻所述视频帧中的所述目标对象是否为同一目标对象，若是，则将相邻所述视频帧中所述目标对象作为同一目标对象进行追踪。
在本申请实施例中，处理器或处理器模块可以是一种集成电路芯片，具有信号的处理能力。处理器可以是通用处理器、数字信号处理器(Digital Signal Processor，简称DSP)、专用集成电路(Application Specific Integrated Circuit，简称ASIC)、现场可编程门阵列(Field Programmable Gate Array，简称FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件。
可以实现或者执行本申请实施例中的公开的各方法、步骤及逻辑框图。通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等。结合本申请实施例所公开的方法的步骤可以直接体现为硬件译码处理器执行完成，或者用译码处理器中的硬件及软件模块组合执行完成。软件模块可以位于随机存储器，闪存、只读存储器，可编程只读存储器或者电可擦写可编程存储器、寄存器等本领域成熟的存储介质中。处理器读取存储介质中的信息，结合其硬件完成上述方法的步骤。
存储介质可以是存储器，例如可以是易失性存储器或非易失性存储器，或可包括易失性和非易失性存储器两者。
其中，非易失性存储器可以是只读存储器(Read-Only Memory，简称ROM)、可编程只读存储器(Programmable ROM，简称PROM)、可擦除可编程只读存储器(Erasable PROM，简称EPROM)、电可擦除可编程只读存储器(Electrically EPROM，简称EEPROM)或闪存。
易失性存储器可以是随机存取存储器(Random Access Memory，简称RAM)，其用作外部高速缓存。通过示例性但不是限制性说明，许多形式的RAM可用，例如静态随机存取存储器(Static RAM，简称SRAM)、动态随机存取存储器(Dynamic RAM，简称DRAM)、同步动态随机存取存储器(Synchronous DRAM，简称SDRAM)、双倍数据速率同步动态随机存取存储器(Double Data Rate SDRAM，简称DDRSDRAM)、增强型同步动态随机存取存储器(Enhanced SDRAM，简称ESDRAM)、同步连接动态随机存取存储器(Synch link DRAM，简称SLDRAM)和直接内存总线随机存取存储器(Direct Ram bus RAM，简称DRRAM)。
本申请实施例描述的存储介质旨在包括但不限于这些和任意其它适合类型的存储器。
本领域技术人员应该可以意识到，在上述一个或多个示例中，本申请所描述的功能可以用硬件与软件组合来实现。当应用软件时，可以将相应功能存储在计算机可读介质中或者作为计算机可读介质上的一个或多个指令或代码进行传输。计算机可读介质包括计算机存储介质和通信介质，其中通信介质包括便于从一个地方向另一个地方传送计算机程序的任何介质。存储介质可以是通用或专用计算机能够存取的任何可用介质。
以上所述的具体实施方式，对本申请的目的、技术方案和有益效果进行了进一步详细说明，所应理解的是，以上所述仅为本申请的具体实施方式而已，并不用于限定本申请的保护范围，凡在本申请的技术方案的基础之上，所做的任何修改、等同替换、改进等，均应包括在本申请的保护范围之内。

Claims (10)

  1. 一种针对运动状态下的多目标对象追踪方法，其特征在于，包括：
    获得视频采集装置中采集的视频数据所包含的视频帧；
    将所述视频帧发送至预设的特征识别模型中，确定所述视频帧中对应目标对象的特征检测区域，从所述特征检测区域中提取所述目标对象的颜色特征，将相邻所述视频帧中所述目标对象的所述颜色特征进行比对，获得第一比对结果；
    确定相邻所述视频帧中所述目标对象的标识部位在目标坐标系中的位置信息，将相邻所述视频帧中的所述标识部位在所述目标坐标系中的位置信息进行比对，获得第二比对结果；
    根据所述第一比对结果和所述第二比对结果，判断相邻所述视频帧中的所述目标对象是否为同一目标对象，若是，则将相邻所述视频帧中所述目标对象作为同一目标对象进行追踪。
  2. 根据权利要求1所述的针对运动状态下的多目标对象追踪方法，其特征在于，所述确定相邻所述视频帧中所述目标对象的标识部位在目标坐标系中的位置信息，具体包括：
    通过预测相邻所述视频帧分别对应的所述视频采集装置的位姿变化情况，获得相邻所述视频帧中分别对应的所述视频采集装置的位姿变化信息；
    根据所述位姿变化信息和相邻所述视频帧中前一视频帧对应的所述视频采集装置的位置信息，确定相邻所述视频帧中后一视频帧对应的所述视频采集装置的位置信息；
    根据相邻所述视频帧中分别对应的所述视频采集装置的位置信息以及所述目标对象的标识部位，利用三角测量法获得所述目标对象的标识部位在以所述视频采集装置为空间坐标原点构建的空间直角坐标系中的位置信息；
    通过坐标变换获得所述目标对象的标识部位在目标坐标系中的位置 信息。
  3. 根据权利要求1所述的针对运动状态下的多目标对象追踪方法，其特征在于，还包括：
    确定所述视频帧中所述目标对象的实际运动区域；
    将所述视频帧中所述目标对象的实际运动区域作为待检测区域，对所述待检测区域之外的所述特征检测区域进行滤除，获得所述待检测区域之内的所述特征检测区域。
  4. 根据权利要求3所述的针对运动状态下的多目标对象追踪方法，其特征在于，所述标识部位为所述目标对象的颈部部位；
    相应的，所述目标对象的标识部位在目标坐标系中的位置信息为所述目标对象的颈部部位位于以所述待检测区域中心为空间坐标原点构建的空间直角坐标系中的位置信息。
  5. 根据权利要求1所述的针对运动状态下的多目标对象追踪方法，其特征在于，所述获得视频采集装置中采集的视频数据所包含的视频帧，具体包括：
    获得所述视频采集装置中采集的所述视频数据，对所述视频数据进行分割处理，获得所述视频数据所包含的视频片段；
    检测所述视频片段之间的特征相似度，将所述特征相似度达到或超过预设相似度阈值并且时间间隔不超过预设时间阈值的视频片段作为一个视频镜头；
    获取所述视频镜头中所包含的视频帧。
  6. 一种针对运动状态下的多目标对象追踪装置，其特征在于，包括：
    视频帧获得单元，用于获得视频采集装置中采集的视频数据所包含的视频帧；
    第一比对单元，用于将所述视频帧发送至预设的特征识别模型中，确定所述视频帧中对应目标对象的特征检测区域，从所述检测区域中提取所述目标对象的颜色特征，将相邻所述视频帧中所述目标对象的所述颜色特征进行比对，获得第一比对结果；
    第二比对单元，用于确定相邻所述视频帧中所述目标对象的标识部位在目标坐标系中的位置信息，将相邻所述视频帧中的所述标识部位在目标坐标系中的位置信息进行比对，获得第二比对结果；
    判断单元，用于根据所述第一比对结果和所述第二比对结果，判断相邻所述视频帧中的所述目标对象是否为同一目标对象，若是，则将相邻所述视频帧中所述目标对象作为同一目标对象进行追踪。
  7. 根据权利要求6所述的针对运动状态下的多目标对象追踪装置，其特征在于，所述确定相邻所述视频帧中所述目标对象的标识部位在目标坐标系中的位置信息，具体包括：
    通过预测相邻所述视频帧分别对应的所述视频采集装置的位姿变化情况，获得相邻所述视频帧中分别对应的所述视频采集装置的位姿变化信息；
    根据所述位姿变化信息和相邻所述视频帧中前一视频帧对应的所述视频采集装置的位置信息，确定相邻所述视频帧中后一视频帧对应的所述视频采集装置的位置信息；
    根据相邻所述视频帧中分别对应的所述视频采集装置的位置信息以及所述目标对象的标识部位，利用三角测量法获得所述目标对象的标识部位在以所述视频采集装置为空间坐标原点构建的空间直角坐标系中的位置信息；
    通过坐标变换获得所述目标对象的标识部位在目标坐标系中的位置信息。
  8. 根据权利要求6所述的针对运动状态下的多目标对象追踪装置，其特征在于，还包括：
    运动区域确定单元，用于确定所述视频帧中所述目标对象的实际运动区域；
    滤除单元，用于将所述视频帧中所述目标对象的实际运动区域作为待检测区域，对所述待检测区域之外的所述特征检测区域进行滤除，获得所述待检测区域之内的所述特征检测区域。
  9. 一种电子设备，其特征在于，包括：
    处理器；以及
    存储器，用于存储针对运动状态下的多目标对象追踪方法的程序，该设备通电并通过所述处理器运行该针对运动状态下的多目标对象追踪方法的程序后，执行下述步骤：
    获得视频采集装置中采集的视频数据所包含的视频帧；
    将所述视频帧发送至预设的特征识别模型中，确定所述视频帧中对应目标对象的特征检测区域，从所述检测区域中提取所述目标对象的颜色特征，将相邻所述视频帧中所述目标对象的所述颜色特征进行比对，获得第一比对结果；
    确定相邻所述视频帧中所述目标对象的标识部位在目标坐标系中的位置信息，将相邻所述视频帧中的所述标识部位在目标坐标系中的位置信息进行比对，获得第二比对结果；
    根据所述第一比对结果和所述第二比对结果，判断相邻所述视频帧中的所述目标对象是否为同一目标对象，若是，则将相邻所述视频帧中所述目标对象作为同一目标对象进行追踪。
  10. 一种存储设备，其特征在于，存储有针对运动状态下的多目标对象追踪方法的程序，该程序被处理器运行，执行下述步骤：
    获得视频采集装置中采集的视频数据所包含的视频帧；
    将所述视频帧发送至预设的特征识别模型中，确定所述视频帧中对应目标对象的特征检测区域，从所述检测区域中提取所述目标对象的颜色特征，将相邻所述视频帧中所述目标对象的所述颜色特征进行比对，获得第一比对结果；
    确定相邻所述视频帧中所述目标对象的标识部位在目标坐标系中的位置信息，将相邻所述视频帧中的所述标识部位在目标坐标系中的位置信息进行比对，获得第二比对结果；
    根据所述第一比对结果和所述第二比对结果，判断相邻所述视频帧中的所述目标对象是否为同一目标对象，若是，则将相邻所述视频帧中 所述目标对象作为同一目标对象进行追踪。
PCT/CN2019/108432 2019-06-17 2019-09-27 一种针对运动状态下的多目标对象追踪方法和装置 WO2020252974A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US17/620,119 US20220215560A1 (en) 2019-06-17 2019-09-27 Method and device for tracking multiple target objects in motion state

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910522911.3A CN110264493B (zh) 2019-06-17 2019-06-17 一种针对运动状态下的多目标对象追踪方法和装置
CN201910522911.3 2019-06-17

Publications (1)

Publication Number Publication Date
WO2020252974A1 true WO2020252974A1 (zh) 2020-12-24

Family

ID=67918698

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/108432 WO2020252974A1 (zh) 2019-06-17 2019-09-27 一种针对运动状态下的多目标对象追踪方法和装置

Country Status (3)

Country Link
US (1) US20220215560A1 (zh)
CN (1) CN110264493B (zh)
WO (1) WO2020252974A1 (zh)

Families Citing this family (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110264493B (zh) * 2019-06-17 2021-06-18 北京影谱科技股份有限公司 一种针对运动状态下的多目标对象追踪方法和装置
KR102311798B1 (ko) * 2019-12-12 2021-10-08 포항공과대학교 산학협력단 다중 객체 추적 방법 및 장치
CN116797971A (zh) * 2019-12-31 2023-09-22 支付宝实验室(新加坡)有限公司 一种视频流识别方法及装置
CN111582036B (zh) * 2020-04-09 2023-03-07 天津大学 可穿戴设备下基于形状和姿态的跨视角人物识别方法
CN111553257A (zh) * 2020-04-26 2020-08-18 上海天诚比集科技有限公司 一种高空抛物预警方法
CN111681264A (zh) * 2020-06-05 2020-09-18 浙江新再灵科技股份有限公司 一种监控场景的实时多目标跟踪方法
CN112101223B (zh) * 2020-09-16 2024-04-12 阿波罗智联(北京)科技有限公司 检测方法、装置、设备和计算机存储介质
CN112991280B (zh) * 2021-03-03 2024-05-28 望知科技(深圳)有限公司 视觉检测方法、系统及电子设备
CN114025183B (zh) * 2021-10-09 2024-05-14 浙江大华技术股份有限公司 直播方法、装置、设备、系统和存储介质
CN115937267B (zh) * 2023-03-03 2023-10-24 北京灵赋生物科技有限公司 一种基于多帧视频的目标轨迹追踪方法
CN116189116B (zh) * 2023-04-24 2024-02-23 江西方兴科技股份有限公司 一种交通状态感知方法及系统

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105405154A (zh) * 2014-09-04 2016-03-16 成都理想境界科技有限公司 基于颜色-结构特征的目标对象跟踪方法
CN105678288A (zh) * 2016-03-04 2016-06-15 北京邮电大学 目标跟踪方法和装置
CN105872477A (zh) * 2016-05-27 2016-08-17 北京旷视科技有限公司 视频监控方法和视频监控系统
CN107025662A (zh) * 2016-01-29 2017-08-08 成都理想境界科技有限公司 一种实现增强现实的方法、服务器、终端及系统
CN110264493A (zh) * 2019-06-17 2019-09-20 北京影谱科技股份有限公司 一种针对运动状态下的多目标对象追踪方法和装置

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1757087A4 (en) * 2004-04-16 2009-08-19 James A Aman AUTOMATIC VIDEO RECORDING OF EVENTS, PURSUIT AND CONTENT PRODUCTION SYSTEM
CN102289948B (zh) * 2011-09-02 2013-06-05 浙江大学 高速公路场景下一种多特征融合的多车辆视频跟踪方法
CN103281477B (zh) * 2013-05-17 2016-05-11 天津大学 基于多级别特征数据关联的多目标视觉跟踪方法
CN105760854B (zh) * 2016-03-11 2019-07-26 联想(北京)有限公司 信息处理方法及电子设备
CN106600631A (zh) * 2016-11-30 2017-04-26 郑州金惠计算机系统工程有限公司 基于多目标跟踪的客流统计方法
KR102022971B1 (ko) * 2017-10-18 2019-09-19 한국전자통신연구원 영상의 객체 처리 방법 및 장치

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105405154A (zh) * 2014-09-04 2016-03-16 成都理想境界科技有限公司 基于颜色-结构特征的目标对象跟踪方法
CN107025662A (zh) * 2016-01-29 2017-08-08 成都理想境界科技有限公司 一种实现增强现实的方法、服务器、终端及系统
CN105678288A (zh) * 2016-03-04 2016-06-15 北京邮电大学 目标跟踪方法和装置
CN105872477A (zh) * 2016-05-27 2016-08-17 北京旷视科技有限公司 视频监控方法和视频监控系统
CN110264493A (zh) * 2019-06-17 2019-09-20 北京影谱科技股份有限公司 一种针对运动状态下的多目标对象追踪方法和装置

Also Published As

Publication number Publication date
CN110264493B (zh) 2021-06-18
CN110264493A (zh) 2019-09-20
US20220215560A1 (en) 2022-07-07

Similar Documents

Publication Publication Date Title
WO2020252974A1 (zh) 一种针对运动状态下的多目标对象追踪方法和装置
WO2021043073A1 (zh) 基于图像识别的城市宠物活动轨迹监测方法及相关设备
CN111462200B (zh) 一种跨视频行人定位追踪方法、系统及设备
CN109145708B (zh) 一种基于rgb和d信息融合的人流量统计方法
US10417503B2 (en) Image processing apparatus and image processing method
WO2018209934A1 (zh) 基于时空约束的跨镜头多目标跟踪方法及装置
WO2020094091A1 (zh) 一种图像抓拍方法、监控相机及监控系统
CN104598883B (zh) 一种多摄像机监控网络中目标再识别的方法
WO2020258978A1 (zh) 对象检测方法和装置
CN108197604A (zh) 基于嵌入式设备的快速人脸定位跟踪方法
WO2020094088A1 (zh) 一种图像抓拍方法、监控相机及监控系统
CN104978567B (zh) 基于场景分类的车辆检测方法
KR101781358B1 (ko) 디지털 영상 내의 얼굴 인식을 통한 개인 식별 시스템 및 방법
WO2019001505A1 (zh) 一种目标特征提取方法、装置及应用系统
CN106991370B (zh) 基于颜色和深度的行人检索方法
CN104361327A (zh) 一种行人检测方法和系统
KR20070016849A (ko) 얼굴 검출과 피부 영역 검출을 적용하여 피부의 선호색변환을 수행하는 방법 및 장치
CN109993086A (zh) 人脸检测方法、装置、系统及终端设备
CN106447701A (zh) 用于图像相似性确定、对象检测和跟踪的方法和装置
CN104573617A (zh) 一种摄像控制方法
US9947106B2 (en) Method and electronic device for object tracking in a light-field capture
Sokolova et al. Human identification by gait from event-based camera
Chen et al. Object tracking over a multiple-camera network
CN109001674B (zh) 一种基于连续视频序列的WiFi指纹信息快速采集与定位方法
TW201530495A (zh) 移動物體追蹤方法及電子裝置

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: 19933715

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: 19933715

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 28.03.2022)

122 Ep: pct application non-entry in european phase

Ref document number: 19933715

Country of ref document: EP

Kind code of ref document: A1