WO2021238710A1 - 人体手部与手势识别的方法及装置、显示设备 - Google Patents

人体手部与手势识别的方法及装置、显示设备 Download PDF

Info

Publication number
WO2021238710A1
WO2021238710A1 PCT/CN2021/094191 CN2021094191W WO2021238710A1 WO 2021238710 A1 WO2021238710 A1 WO 2021238710A1 CN 2021094191 W CN2021094191 W CN 2021094191W WO 2021238710 A1 WO2021238710 A1 WO 2021238710A1
Authority
WO
WIPO (PCT)
Prior art keywords
target
detection
human
human body
frame
Prior art date
Application number
PCT/CN2021/094191
Other languages
English (en)
French (fr)
Inventor
刘成
郭思佳
郭冠出
Original Assignee
京东方科技集团股份有限公司
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 京东方科技集团股份有限公司 filed Critical 京东方科技集团股份有限公司
Priority to US17/790,416 priority Critical patent/US11797098B2/en
Publication of WO2021238710A1 publication Critical patent/WO2021238710A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
    • G06F3/017—Gesture based interaction, e.g. based on a set of recognized hand gestures
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
    • G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/88—Radar or analogous systems specially adapted for specific applications
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/88—Radar or analogous systems specially adapted for specific applications
    • G01S13/89—Radar or analogous systems specially adapted for specific applications for mapping or imaging
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
    • G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
    • G01S7/415—Identification of targets based on measurements of movement associated with the target
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/20—Movements or behaviour, e.g. gesture recognition
    • G06V40/23—Recognition of whole body movements, e.g. for sport training
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/20—Movements or behaviour, e.g. gesture recognition
    • G06V40/28—Recognition of hand or arm movements, e.g. recognition of deaf sign language

Definitions

  • the present disclosure relates to the field of radar detection technology, for example, to a method and device for human hand and gesture recognition, and a display device.
  • Millimeter wave radar refers to a radar whose working frequency is in the millimeter wave frequency band. Since the wavelength of millimeter wave is between centimeter wave and light wave, millimeter wave has the advantages of microwave guidance and photoelectric guidance. Compared with infrared, laser, TV and other optical seekers, the millimeter waveguide seeker has a strong ability to penetrate fog, smoke, and dust, and has the characteristics of all-weather all-weather. Millimeter wave radar is a high-precision sensor that measures the relative distance, relative speed, and azimuth of the measured object. It was used in the military field in the early days. With the development and progress of radar technology, millimeter wave radar sensors have begun to be used in automotive electronics, unmanned aerial vehicles, Intelligent transportation and other fields.
  • a method for human hand recognition includes: identifying a human target using multiple frames of detection information collected by a millimeter-wave radar within a preset time period; and determining whether there is a human target within a preset range based on the detection information of the current frame, centering on the human target Conditional new detection target, the setting conditions include: having a radial velocity; if it exists, the new detection target that meets the setting conditions is determined as the hand of the human target; if it does not exist, it is determined that the current There is no hand of the human target in the frame.
  • using the multi-frame detection information collected by the millimeter-wave radar within a preset time period to identify a human target includes: generating and maintaining a reference object map based on the detection target in the multi-frame detection information, and using all The reference object map identifies the target detected by the millimeter wave radar as a stationary object or a non-stationary object; generates and maintains a human body map based on the non-stationary object identified by the reference object map, and uses the human body map to detect from the multiple frames Identify the human target among the non-stationary objects in the information.
  • the generating and maintaining a reference object map based on the detection target in the multi-frame detection information includes: generating and maintaining the reference object map based on the parameters of the detection target in the multi-frame detection information, the The parameters include at least one of spatial coordinates, radial velocity, signal-to-noise ratio, noise, heat of presence, heat of recent movement, and the number of consecutive detections.
  • the presence heat is used to indicate the frequency with which the detection target is detected in multiple frames of detection information.
  • the higher the frequency the greater the presence heat; and/or, the recent movement heat is used to indicate The time interval between the last frame in which the detection target is detected in motion and the current frame.
  • the smaller the time interval the greater the value of the recent motion popularity; and/or, the number of consecutive detections is used to indicate that the detection target is continuously detected.
  • the maintenance of the reference object map includes: extracting the detection targets in each frame of the multi-frame detection information one by one according to the detection order; The detection target calculates the spatial distance from the existing target in the reference object map one by one; if the spatial distance is less than the first threshold, it is determined that the detection target and the existing target with the smallest spatial distance are the same measured target; if the space is The distance is not less than the first threshold, then a measured target is created for the detection target in the reference object image.
  • the method further includes: updating the parameters of the existing target: the spatial coordinates are updated to the difference between the detection target and the existing target The average value of the space coordinates; the radial velocity, signal-to-noise ratio and noise are updated to the parameter values corresponding to the detection target; the presence of heat is increased in the preset range based on the gradient value corresponding to the number of frames; the recent movement is heated when the radial velocity is available Update to the maximum value of recent sports heat; the number of consecutive detections is updated based on the detection status of the previous frame within the preset range; if the number of consecutive detections reaches the maximum, the presence heat is set to the maximum value, and the recent sports heat is set to 0.
  • the method further includes: marking the existing target as a stationary object if the presence heat of the existing target is higher than a preset value and the recent motion heat is 0; and /Or, if the number of consecutive detections of the existing target reaches a preset minimum value, it is marked as a non-stationary object.
  • the method further includes: determining the parameters of the created measured target: wherein, the spatial coordinates, the radial velocity, the signal-to-noise ratio and the The value of the noise is the parameter value corresponding to the detection target; the presence heat is 1; the recent movement heat is set to the maximum value when it has a radial velocity, otherwise it is 0; the number of consecutive detections is set to 1.
  • the maintenance of the reference object map further includes: if the time point corresponding to the current frame is the end time point of the initialization process of the millimeter wave radar, updating all measured targets in the reference object map Parameter: Decrease the heat of each measured target within a preset range; if the existing target in the reference object map does not have a corresponding measured target in the current frame, then continuous detection of the existing target When the number of times is a positive value, set it to -1.
  • the initialization process is a process in which the millimeter-wave radar starts when the first detection frame appears and ends when the preset time is reached, and processes the collected multi-frame detection information to generate and maintain a reference object map.
  • the maintenance of the human body map includes: acquiring a detection target of one frame of detection information in the multi-frame detection information; if the detection target is marked as a non-stationary object in the reference object map, Then the detection target is determined as a human target.
  • the maintenance of the human body map includes: acquiring a detection target of one frame of detection information in the multi-frame detection information; if the detection target is marked as a non-stationary object in the reference object map, In addition, if the detection target has other detection targets that are also marked as non-stationary objects in the designated range on the human body map, the detection target and the other detection targets are jointly determined as the human body target.
  • the method further includes: judging whether the planar projection distance between the human body target and the existing target in the human body map is less than a second threshold; The target and the existing target are merged into a new human body target, and the parameters of the new human body target in the human body map are updated; if not less than, a new human body target corresponding to the human body target is created in the human body map, And determine the parameters of the new human target according to the parameters of the human target.
  • the maintenance of the human body map further includes: comparing the human body target determined from the one frame of detection information with the human body target determined jointly by multiple frames of detection information before the one frame of detection information; if If the plane projection distance of the compared human target in the human body map is less than the third threshold, the human target determined by the multiple frames of detection information is merged, and the parameters of the merged human target in the human body map are updated.
  • the determining the new detection target that meets the set condition as the hand of the human target further includes: if there are multiple new detection targets that meet the set condition, then according to the millimeter wave radar and The distance of each new detection target that meets the set conditions determines the new detection target corresponding to the hand of the human target.
  • the setting condition further includes: the echo signal strength of the signal-to-noise ratio is lower than the human body target.
  • the method for human gesture recognition includes: determining the spatial position and radial velocity of the hand target in a plurality of consecutive frames by using the hand target corresponding to the human target detected by the millimeter wave radar; Determine the motion trajectory of the hand target when the motion trajectory reaches a preset length; perform a straight line fitting on the motion trajectory; determine the human body according to the direction vector of the straight line obtained by the fitting and the radial velocity Human hand gesture for target.
  • the straight line fitting includes: spatial straight line fitting; and/or, projecting the motion trajectory onto a preset plane, and performing straight line fitting in the preset plane.
  • the determining the motion trajectory of the hand target according to the spatial position and the radial velocity includes: determining when the hand target is not detected in the multiple consecutive frames Whether the number of consecutive frames where the hand target is not detected is greater than the preset number of frames; if the number of consecutive frames where the hand target is not detected is greater than the preset number of frames, stop determining the motion trajectory of the hand target .
  • a device for human hand recognition includes: an identification unit, a judgment unit, and a determination unit.
  • the recognition unit is configured to recognize the human target by using multiple frames of detection information collected by the millimeter wave radar within a preset time period.
  • the judging unit is configured to determine whether there is a new detection target that satisfies a set condition within a preset range based on the detection information of the current frame with the human target determined by the identification unit as the center; the set condition includes: speed.
  • the determining unit is configured to, if the determining unit determines that there is a new detection target that meets the set conditions, determine it as the human target hand; and also configured to determine that there is no hand that meets the set conditions. If the condition is new to detect the target, it is determined that there is no hand of the human target in the current frame.
  • a device for human gesture recognition includes a first determination unit, a second determination unit, a fitting unit and a gesture recognition unit.
  • the first determining unit is configured to use the hand target corresponding to the human target detected by the millimeter wave radar to determine the spatial position and radial velocity of the hand target in a plurality of consecutive frames.
  • the second determining unit is configured to determine the motion trajectory of the hand target according to the spatial position and the radial velocity determined by the first determining unit.
  • the fitting unit is configured to perform spatial straight line fitting on the movement trajectory when the movement trajectory determined by the second determination unit reaches a preset length.
  • the gesture recognition unit is configured to determine the human body gesture of the human target according to the direction vector of the spatial straight line and the radial velocity obtained by the fitting unit.
  • a processor for running a program that executes the human hand recognition method described in any of the above embodiments or executes the human hand recognition method described in any of the above embodiments when the program is running.
  • the method of gesture recognition is provided for running a program that executes the human hand recognition method described in any of the above embodiments or executes the human hand recognition method described in any of the above embodiments when the program is running. The method of gesture recognition.
  • a display device in another aspect, includes: a display panel; a millimeter wave radar integrated in the display panel, the millimeter wave radar is configured to collect multiple frames of detection information within a preset time period; and as described in any of the above embodiments The processor described.
  • a computer-readable storage medium stores computer program instructions, and when the computer program instructions run on a processor, the processor executes one of the human hand recognition methods described in any of the above embodiments. Or multiple steps; or perform one or more steps in the method for human gesture recognition described in any of the foregoing embodiments.
  • a computer program product in another aspect, includes computer program instructions, and when the computer program instructions are executed on a computer, the computer program instructions cause the computer to execute one or more of the human hand recognition methods described in any of the above embodiments. Steps; or perform one or more steps in the method for human gesture recognition as described in any of the above embodiments.
  • a computer program When the computer program is executed on a computer, the computer program causes the computer to execute one or more steps in the method for human gesture recognition as described in any of the above embodiments; or executes as in any of the above embodiments One or more steps in the method of human gesture recognition described in.
  • Fig. 1 is a flowchart of a method for human hand recognition according to some embodiments
  • FIG. 2 is a flowchart of a method for identifying a human target using multiple frames of detection information collected by millimeter wave radar within a preset time period according to some embodiments;
  • Fig. 3 is a flowchart of a method for maintaining a reference object map according to some embodiments
  • Fig. 4 is a flowchart of another method of maintaining a reference object map according to some embodiments.
  • Fig. 5 is a flowchart of a method for maintaining a human body map according to some embodiments
  • Fig. 6 is a flowchart of another method for maintaining a human body map according to some embodiments.
  • Fig. 7 is a flowchart of a method for human gesture recognition according to some embodiments.
  • Fig. 8 is a flowchart of a method for determining the motion trajectory of the hand target according to the spatial position and radial velocity according to some embodiments;
  • Fig. 9 is a structural diagram of a device for human hand recognition according to some embodiments.
  • FIG. 10 is a structural diagram of another device for human hand recognition according to some embodiments.
  • FIG. 11 is a structural diagram of a device for human gesture recognition according to some embodiments.
  • Fig. 12 is a structural diagram of another device for human gesture recognition according to some embodiments.
  • Fig. 13 is a structural diagram of a display device according to some embodiments.
  • first and second are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined with “first” and “second” may explicitly or implicitly include one or more of these features.
  • plural means two or more.
  • At least one of A, B, and C has the same meaning as “at least one of A, B, or C", and both include the following combinations of A, B, and C: only A, only B, only C, A and B The combination of A and C, the combination of B and C, and the combination of A, B and C.
  • a and/or B includes the following three combinations: A only, B only, and the combination of A and B.
  • the term “if” is optionally interpreted to mean “when” or “when” or “in response to determination” or “in response to detection.”
  • the phrase “if it is determined" or “if [the stated condition or event] is detected” is optionally interpreted to mean “when determining" or “in response to determining" Or “when [stated condition or event] is detected” or “in response to detecting [stated condition or event]”.
  • Millimeter-wave radar has certain advantages when detecting moving targets. For example, it can be used for human detection or hand detection. Among them, whether it is a human target or a hand target, it is because of its interaction with millimeter wave. The relative movement of the radar can be accurately identified and detected. However, when the human body and its hands are in motion at the same time, because the distance between the two is relatively close, the motions of the two will interfere with each other, making it impossible to effectively distinguish the human body or the hand in motion, and it is impossible to recognize the hand movement. Out gesture.
  • some embodiments of the present disclosure provide a method for human hand recognition, which can recognize and detect the human body and its hands at the same time. As shown in FIG. 1, the method includes S1 to S4.
  • This step is based on the analysis of the multi-frame detection information of the millimeter wave radar, and the human target is determined by comparing the measured targets in different frames.
  • multiple parameters need to be set for each measured target, including conventional parameters and custom parameters, such as space coordinates, radial velocity (that is, the measured target and millimeter wave radar The rate of change of the space distance between each other), signal-to-noise ratio and noise, etc., and the custom parameters are mainly used to describe the motion state of a measured target in multiple consecutive frames to assist in judging whether the test target is a stationary object or a non-stationary object. Stationary object, and further determine whether it is a human target.
  • the non-stationary objects here include objects that move within the detection range and objects that move out of the detection range.
  • the detection information of the current frame it is determined whether there is a new detection target that meets the set condition within the preset range with the human target as the center; the set condition includes: having a radial velocity.
  • the detection information of the current frame is used as the content of the analysis, because in the detection information of multiple frames, the position of the same human target may change, resulting in the inaccuracy of the preset range determined with the human target as the center. Therefore, the new detection target cannot be effectively determined.
  • the measured target detected by the millimeter-wave radar is a target with spatial coordinates.
  • the human hand is between the human body and the millimeter-wave radar, although the hand may block part of the human body, it is analyzed by the position of the spatial coordinate. It is still possible to distinguish human targets from hand targets.
  • the new detection target in this step is a target that appears near the human target in the current frame and can be detected by the millimeter wave radar.
  • the current frame in this step is one frame of detection information in the multiple frames of detection information in S1; the radial velocity in this step is used to indicate the rate of change of the spatial distance between the newly detected target and the millimeter wave radar.
  • the setting conditions also include: the signal-to-noise ratio is lower than the signal-to-noise ratio of the human target.
  • the new detection target can be determined as the candidate hand for the human target, and S3 is executed at this time; on the contrary, when there is no new detection target that meets the set conditions, then Perform S4.
  • gestures can be further recognized.
  • the new detection target does not meet the above set conditions, it can be determined that it is not a hand; or it is determined that it is a hand but the gesture to be recognized is not made. Recognition.
  • the new detection target is determined as the hand of the human target according to the distance between the millimeter wave radar and each new detection target that meets the set conditions.
  • the new detection target that is closest to the millimeter-wave radar and meets the set conditions is determined as the hand of the human target.
  • the new detection target that is closest to the millimeter wave radar and meets the set conditions is determined as the human target hand is only a selection based on the general application scenario of the above steps, because the human body is usually It faces the millimeter wave radar, stretches out the hand forward and makes an action.
  • the hand of the human target is usually the closest to the millimeter wave radar. That is to say, in other application scenarios, when determining which new detection target is the hand of the human target based on the distance, other basis, such as the farthest distance, can also be used for judgment.
  • the method of human hand recognition based on millimeter wave radar adopted in each embodiment of the present disclosure is to analyze multiple frames of detection information collected by millimeter wave radar within a preset time period. Use custom parameters to assist in determining whether there is a human target in the measured target, and analyze the new detection target suspected to be a hand based on the detection information of the current frame. When it is determined that there is a new detection target, it is based on whether the new detection target meets the setting Condition to identify whether it is the hand of the human target. In this way, through the above steps, the human body and its hand within the detection range can be effectively identified through the detection of the millimeter wave radar, so as to facilitate subsequent gesture recognition for the hand.
  • the multi-frame detection information collected by the millimeter wave radar in the preset time period in S1 is used to identify the human target, as shown in FIG. 2, including S11 to S12.
  • the reference object map is generated and maintained based on the parameters of the detection target in the multi-frame detection information.
  • the reference object map in this step is generated based on the detection target in the first frame of detection information, and can be maintained based on the detection target in the subsequent multi-frame detection information. That is, before processing the first frame of detection information, there may be no reference object image, and the reference object image is generated based on the detection target obtained from the first frame of detection information; when processing the second frame of detection information, the first frame of detection information The detection target becomes the existing target, and the distance comparison with the detection target in the second frame of detection information can update the existing target in the reference object map to realize the maintenance of the reference object map.
  • new detection targets can be continuously added to the reference object map.
  • the parameters of the detection target include conventional parameters, such as spatial coordinates, radial velocity, signal-to-noise ratio, and noise.
  • the parameters of the detection target include custom parameters, such as presence heat, recent sports heat, and the number of consecutive detections.
  • the presence of heat is used to indicate the frequency with which the detection target is detected in the multi-frame detection information. The higher the frequency, the greater the heat; the recent movement heat is used to indicate the last frame of the detection target's movement and the current.
  • the time interval between frames the smaller the time interval, the greater the value of the recent sports heat; the number of consecutive detections, used to indicate the number of times that the detection target is continuously detected or continuously undetected, where a positive value indicates that it has been continuously detected , Negative value means continuous undetected.
  • the human body map is similar to the reference object map, is generated based on the detection information of the first frame, and can be maintained based on the subsequent multi-frame detection information. That is, before processing the first frame of detection information, there may be no human body map, and generate a human body map based on the first frame of detection information; when processing the second frame of detection information, the non-stationary object of the first frame of detection information becomes already If there is a target, the distance comparison with the non-stationary object target in the second frame of detection information can update the human body target in the human body map to realize the maintenance of the human body map.
  • the difference between the human body map and the reference object map is that the reference object map is a map in a three-dimensional space coordinate system; while the human body map is a two-dimensional map, which can be the ground in the application scene, or the plane of the human body map can be set according to requirements.
  • the parameters of the human body target recorded in the human body map include: plane coordinates, height (human body height), recent exercise fever, and cumulative target number. Among them, the cumulative target number refers to the target points detected by the radar.
  • the maintenance reference object map in S11 includes S111 to S116.
  • S111 Extract the detection target in each frame of the multi-frame detection information one by one according to the detection order.
  • This step is to extract the data of the detection target frame by frame according to the number of frames collected, including the specific parameters of the detection target; among them, by obtaining the space coordinates of the detection target and the value of the radial velocity, it can be judged whether the detection target needs subsequent steps.
  • the analysis that is to say, judge the rationality of these detection targets, and determine that the detection target is a real object.
  • the basis for judging the rationality of these detection targets is whether the distance between the detection target and the millimeter wave radar is within a preset detection range; wherein, the preset detection range is set according to actual application requirements.
  • S112 Calculate the spatial distances of the detection targets in one frame of detection information from the existing targets in the reference object map one by one according to the spatial position of the detection target.
  • each detection target in each frame of detection information is compared with the reference object map one by one, and the spatial distance between it and the existing target is calculated, so as to determine whether the detection target is an existing target in the reference object map.
  • the detection target detected by the millimeter wave radar can be represented as a point in the reference object map, and the size of the point can be determined based on the resolution of the radar, or can be set artificially according to the resolution. In this way, for a human target, its representation in the reference object map will become a collection of multiple detection targets.
  • the two detection targets are regarded as the same measured target.
  • the method of updating the parameters of the existing target may include: updating the space coordinates to the average value of the space coordinates of the detected target and the existing target (that is, updating the coordinate position of the object in the figure); radial velocity, signal noise The ratio and noise are updated as the parameter value corresponding to the detection target; the presence of heat increases within the preset range based on the gradient value corresponding to the number of frames (for example, the preset range is 0-100, and the gradient value is 1, then when it is determined that the detection target is Have
  • the target is the same target, the value of the existing heat is increased by 1 until it reaches the maximum value and no longer increases); the recent sports heat is updated to the maximum value of the recent sports heat when the radial velocity is available (the recent sports heat is also a range value , When the detection target exists and moves in the multi-frame detection information, the value of the recent motion heat is greater, and when the detection target stops moving, the value of the recent motion heat is reduced); the number of consecutive detections is preset The detection status is updated based on the previous frame in the range.
  • the update is determined based on the value of the previous frame (the value of the existing target) Yes, when the value of the previous frame is positive (continuously detected), the value is increased by 1, and when the value of the previous frame is negative (continuously undetected), the value is set to 1, and the number of consecutive detections is also A range value (the number of frames of multi-frame detection information), generally from 1 to the maximum value (continuously detected) or -1 to the minimum value (continuously undetected).
  • the heat of the detection target's existence is set to the maximum, and the heat of the recent exercise is set to 0.
  • the updated existing target's presence heat is higher than the preset value and the recent motion fever is 0, it means that the object corresponding to the existing target can be detected all the time and no movement occurs. At this time, mark The existing target is a stationary object.
  • the preset value of heat can be manually set.
  • the existing target if the number of consecutive detections of the updated existing target reaches the preset minimum value, it means that the existing target once appeared in the reference object map, but it cannot be in the corresponding position for a long period of time.
  • the existing target is detected, that is, the existing target has moved, and the existing target is marked as a non-stationary object at this time.
  • the space distance between the detection target and multiple existing targets at the same time is less than the first threshold, merge it with the nearest existing target and update the parameter value at the same time; or The existing target and the detection target are all merged into one target, and then the parameters of the target are updated.
  • the method for determining the parameters of the measured target may include: spatial coordinates, radial velocity, signal-to-noise ratio and noise values are the parameter values corresponding to the detected target; the presence degree is set to 1; The radial speed is set to the maximum value, otherwise it is 0; the number of consecutive detections is set to 1.
  • the first threshold in some of the foregoing embodiments may be the resolution of the radar, or may be a value manually set. This disclosure does not limit this.
  • the maintenance reference object map in S11 includes S111 to S116 in the foregoing embodiment, and also includes S117.
  • the initialization process of this step is a process in which the millimeter-wave radar starts when the first detection frame appears and ends when the preset time is reached, and processes the collected multi-frame detection information to generate a reference object map. That is, the initialization process is a process of processing the detection information by the millimeter wave radar with a time length artificially set.
  • the purpose of the initialization process is to quickly determine which detection targets in the detection range are stationary objects after the millimeter-wave radar starts to work, because a detection target generally needs to be detected for a long time to determine that it is a stationary object, and this step
  • the initialization process in can forcibly reduce the detection time. If the movable object does not move during the initialization process, then the object is also marked as a stationary object until the object moves in the subsequent detection process. It will be marked as a non-stationary object.
  • the method of updating the parameters of all the measured targets in the reference object map may include:
  • the preset range is from 0 to a preset value, and the decrease process is also decreased according to the preset gradient until it decreases to 0, that is After each frame of data is processed, the measured target's existence enthusiasm is updated once.
  • the existing target in the reference object image does not have a corresponding detection target in the frame (that is, when there is no detection target combined with the existing target in the frame), then when the number of consecutive detections is positive, Set it to -1; when the number of consecutive detections is negative, decrease the value by 1.
  • Mark the measured target that has a heat higher than the preset value and the recent motion heat is 0 as a stationary object; mark the measured target whose number of consecutive detections reaches the preset minimum value (that is, the number of consecutive frames has not been detected) Stationary object; delete the measured target with a heat degree of 0 from the reference object map, that is, delete the object that has appeared or occasionally appears.
  • the maintenance of the human body map in S12 includes S121 to S125.
  • the detection target marked as a non-stationary object in the reference object map is determined as a human target.
  • S123 Determine whether the plane projection distance between the human body target and the existing target in the human body map is less than a second threshold. This step is to project the human body target in the frame onto the plane where the human body map is located, and determine the distance between it and the existing target in the plane. If the plane projection distance is less than the second threshold, execute S124; otherwise, If the plane projection distance is not less than the second threshold, S125 is executed.
  • the second threshold is generally set according to the area occupied by a person in the projection of the human body map, such as 0.5m. When it is less than the second threshold, it is considered that the human target and the existing target are the same human body. Of different parts.
  • the existing target in the human body map refers to the target obtained by the accumulation of multiple frames of detection information processed before the frame.
  • the specific method for updating the parameters of the new human target includes: updating the plane coordinates to a weighted average (that is, the plane coordinates of the existing target are weighted by the cumulative target number, and the plane coordinates of the human target are weighted by 1); The cumulative target number is increased by 1; the height is updated to the larger of the current value of the target in the human body map and the height of the human target; the recent movement heat is updated to the comparison between the current value of the human target in the human body map and the recent movement heat of the existing target The bigger one.
  • a weighted average that is, the plane coordinates of the existing target are weighted by the cumulative target number, and the plane coordinates of the human target are weighted by 1
  • the cumulative target number is increased by 1
  • the height is updated to the larger of the current value of the target in the human body map and the height of the human target
  • the recent movement heat is updated to the comparison between the current value of the human target in the human body map and the recent movement heat of the existing target The bigger one.
  • This step is to create a new human body target in the human body map when the distance between the human body target in the current frame and the existing target in the human body map is not close enough, that is, when it is not less than the second threshold.
  • Target setting parameters
  • the specific method for setting parameters for the new human body target includes: plane coordinates, height, and recent sports heat are all parameters of the human body target, and the cumulative target number is set to 1.
  • the maintenance of the human body map in S12, as shown in FIG. 6, is replaced by S122' on the basis of S121 to S125 in the foregoing embodiment.
  • the detection target is marked as a non-stationary object in the reference object map, and there are other detection targets that are also marked as non-stationary objects in the range specified in the human body map, the detection target is combined with other detection targets Jointly determine as the human body target, thereby reducing the analysis of the human body target one by one, and improving the processing efficiency.
  • the designated range may be the range occupied by the projection of the human body on the human body map.
  • the maintenance of the human body map in S12 further includes S126.
  • Some of the above embodiments update and maintain the human body map for the detection target in one frame of detection information. After traversing all the detection targets in one frame of detection information, traverse the human body determined in the previous multiple frames of detection information in units of frames.
  • Target if the plane projection distance of the human target on the human body map is less than the third threshold, the third threshold is a custom distance, that is, to determine whether the human target determined in different frames is close enough, and multi-frame detection information (including the current frame and The previous multi-frame detection information) is determined to be close enough to the human body target to be merged, and the parameters of the merged human target in the human body map are updated.
  • the specific human body target can be determined, and the recent movement fever of the general human body target is not zero.
  • the height of the human body detected by the millimeter wave radar can also be determined, and the detected person can be further distinguished as an adult or a child based on the height.
  • Some embodiments of the present disclosure provide a method for human body gesture recognition, which is a method for quickly recognizing the gesture of the hand to determine the human body gesture after the human body and the corresponding hand are recognized at the same time, as shown in FIG. 7 ,
  • the method includes S100 ⁇ S400.
  • the hand targets in this step have corresponding human targets, when multiple human targets and multiple hand targets are recognized, it can be more accurately judged which person made which gestures, and the feedback is more accurate Gesture response result.
  • this step is based on a frame as a unit to obtain the spatial position and radial velocity of the hand target in multiple consecutive frames.
  • the hand target corresponding to the human target detected by the millimeter wave radar in S100 determines the spatial position and radial velocity of the hand target in multiple consecutive frames, including S101 to S102.
  • the preset length is the number of hand targets in different spatial positions, and this number has hand targets in multiple consecutive frames, and the spatial position of each hand target is not the same as that of the frame.
  • the numbers are the same.
  • the hand target is not detected in a certain frame due to interference or other problems in multiple consecutive frames, or the hand target movement is slow, resulting in the space of the hand target in multiple consecutive frames If the position is the same, at this time, it is necessary to collect the spatial position and radial velocity of the hand target in more frames to form a motion trajectory to make it conform to the preset length.
  • the setting of the preset length is also related to the difficulty of the gesture to be recognized, that is, simple gestures can be recognized through a shorter-length motion trajectory, while complex gestures require a longer motion trajectory to be recognized.
  • S102 Acquire the spatial position and radial velocity of the hand target in multiple consecutive frames according to the preset length.
  • This step is to determine the number of frames of multiple consecutive frames based on the preset length, and then obtain the spatial position and radial velocity of the hand target from the multiple consecutive frames.
  • S200 Determine the movement trajectory of the hand target according to the spatial position and the radial velocity.
  • multiple spatial points can be determined according to the spatial position of the hand target, and the movement direction of the hand target can be determined according to the radial velocity corresponding to each spatial position, and these spatial points are connected with the movement direction as a reference to obtain The trajectory of the hand target.
  • the motion trajectory in this step is determined based on the detection information of multiple frames.
  • the length of the motion trajectory is positively related to the number of frames.
  • the preset length refers to the movement trajectory of the hand target determined by the detection information of a specified number of consecutive frames.
  • a spatial straight line fitting is performed on the motion trajectory.
  • the least square method may be used to fit the motion trajectory to a straight line in space.
  • the motion trajectory when the motion trajectory reaches a preset length, the motion trajectory is projected onto a preset plane, and a straight line is fitted in the preset plane.
  • the spatial straight line fitting there are generally two results: one is the high linear correlation between the motion trajectory and the spatial straight line fitting, that is, it can be fitted to a spatial straight line.
  • the fitted spatial straight line further judges specific human gestures.
  • the other is that the linear correlation between the motion trajectory and the spatial straight line fitting is not high, that is, it cannot be fitted to a spatial straight line.
  • the motion trajectory is projected into a preset plane, and the process is performed in the preset plane.
  • Straight-line fitting among them, the preset plane is preset according to actual requirements, such as the plane where the transceiver antenna of the millimeter wave radar is located.
  • the human body gesture recognition is not performed on the motion trajectory.
  • S400 Determine the human body gesture of the human target according to the direction vector and the radial velocity of the straight line obtained by the fitting.
  • the human body gestures recognized in this step include, but are not limited to, left and right, up and down, and forward and backward sliding gestures made by the human body facing the millimeter wave radar.
  • the left and right, up and down sliding gestures can be recognized by the direction vector of the spatial straight line
  • the forward and backward sliding gestures can be recognized by the radial velocity of the hand target. Combining the direction vector of the straight line and the radial velocity can recognize more sliding gestures.
  • the multiple consecutive frames corresponding to the motion trajectory are updated, that is, the data points that constitute the motion trajectory are updated. Update.
  • updating the multiple consecutive frames corresponding to the motion trajectory includes:
  • the specific human gesture is determined by shortening the motion trajectory, so that additional non-gesture actions of the hand before the gesture action can be removed.
  • the recognition of the human gesture for the human target is suspended for a preset period of time.
  • This step is to prevent the human body from being recognized as a new gesture when the hand moves to the next point in preparation for the second gesture after the human body makes a gesture.
  • the foregoing embodiment is based on the simultaneous recognition of the human body and the hand to obtain the hand target corresponding to the human body target, and realizes the tracking of the movement trajectory of the hand target, so as to quickly recognize the specific gesture made by the human body.
  • determining the movement trajectory of the hand target according to the spatial position and radial velocity in S200 includes S201 and S202.
  • S201 When the hand target is not detected in the multiple consecutive frames, determine whether the number of consecutive frames in which the hand target is not detected is greater than the number of frames.
  • the above steps allow for multiple consecutive frames where the hand target is not detected in individual frames.
  • the number of consecutive frames where the hand target is not detected needs to be limited, because when the number of consecutive frames is greater than the preset frame After counting, it can be considered that the hand target has finished the gesture, or the gesture has been interrupted. At this time, the determination of the movement trajectory of the hand target should be stopped.
  • Some embodiments of the present disclosure provide a device for human hand recognition, which can simultaneously and accurately recognize the human body and its hand target within the measured range by using millimeter wave radar.
  • This embodiment corresponds to the foregoing method embodiment.
  • this embodiment will not repeat the details in the foregoing method embodiment one by one, but it should be clear that the device in this embodiment can correspondingly implement the foregoing method embodiment.
  • the device 6 includes: an identification unit 61, a judgment unit 62 and a determination unit 63.
  • the recognition unit 61 is configured to recognize a human target by using multiple frames of detection information collected by the millimeter wave radar within a preset time period.
  • the judging unit 62 is configured to determine whether there is a new detection target that meets a set condition within a preset range based on the detection information of the current frame and centering on the human target determined by the recognition unit 61, and the set condition includes: To speed.
  • the determining unit 63 is configured to, if the determining unit 62 determines that there is a new detection target that meets the set conditions, determine the new detection target that meets the set conditions as the hand of the human target; and is also configured to The determining unit 62 determines that there is no new detection target that meets the set conditions, and then determines that there is no hand of the human target in the current frame.
  • the above-mentioned setting condition further includes: the signal-to-noise ratio is lower than the signal-to-noise ratio of the human target.
  • the determining unit 63 is further configured to determine the hand of the human target according to the distance between the millimeter wave radar and each new detection target that meets the set condition when it is determined that there are multiple new detection targets that meet the set conditions. The corresponding new detection target.
  • the determining unit 63 is further configured to determine the new detection target closest to the millimeter wave radar that satisfies the set conditions as the hand of the human target within the measurement range of the millimeter wave radar.
  • the identification unit 61 includes a first identification module 611 and a second identification module 612.
  • the first identification module 611 is configured to generate and maintain a reference object map based on the detection target in the multi-frame detection information, and use the reference object map to identify the target detected by the millimeter wave radar as a stationary object or a non-stationary object.
  • the second recognition module 612 is configured to generate and maintain a human body map based on the non-stationary objects recognized by the reference object map obtained by the first recognition module 611, and use the human body map to determine the human body target from the non-stationary objects in the multi-frame detection information.
  • the first recognition module 611 is configured to generate and maintain a reference object map based on the parameters of the detection target in the multi-frame detection information, and use the reference object map to identify the target detected by the millimeter wave radar as a stationary object or a non-stationary object. object.
  • the parameters of the detection target include: at least one of spatial coordinates, radial velocity, signal-to-noise ratio, noise, presence heat, recent motion heat, and the number of consecutive detections.
  • Existence heat is used to indicate the frequency with which the detection target is detected in multiple frames of detection information. The higher the frequency, the greater the heat.
  • the recent motion fever is used to indicate the time interval between the last frame in which the detection target was detected motion and the current frame. The smaller the time interval, the greater the recent motion fever value.
  • the number of consecutive detections is used to indicate the number of times that the detection target is continuously detected or continuously undetected. A positive value indicates that it has been continuously detected, and a negative value indicates that it has not been continuously detected.
  • the first identification module 611 includes an extraction submodule 6111, a calculation submodule 6112, a merging submodule 6113, and a creation submodule 6114.
  • the extraction submodule 6111 is configured to extract the detection targets in each frame of the multi-frame detection information one by one according to the detection sequence.
  • the calculation sub-module 6112 is configured to calculate the spatial distance between the detection target in a frame of detection information and the existing target in the reference object map one by one according to the spatial position of the detection target obtained by the extraction sub-module 6111.
  • the merging sub-module 6113 is configured to determine that the detection target and the existing target with the smallest spatial distance are the same measured target if the spatial distance obtained by the calculation sub-module 6112 is less than the threshold; After the target is the same measured target, update the parameters of the existing target.
  • the creation sub-module 6114 is configured to create a measured target for the detection target in the reference object image if the spatial distance obtained by the calculation sub-module 6112 is not less than the threshold; and is also configured to be the detection target in the reference object image After creating the measured target, determine the parameters of the created measured target.
  • the specific method of updating the parameters of an existing target includes: updating the spatial coordinates to the average value of the detection target and the existing target's spatial coordinates; updating the radial velocity, signal-to-noise ratio and noise to the parameter values corresponding to the detection target; Existence heat rises based on the gradient value corresponding to the number of frames within the preset range; the latest sports heat is updated to the maximum value of the most recent sports heat when it has a radial speed; the number of consecutive detections is within the preset range based on the detection status of the previous frame Update; among them, if the number of consecutive detections reaches the maximum, the presence heat is set to the maximum, and the recent exercise heat is set to 0.
  • the existing target is marked as a stationary object; if the number of consecutive detections of the existing target reaches the preset minimum value, it is marked as a non-stationary object.
  • the specific method for determining the parameters of the created measured target includes: spatial coordinates, radial velocity, signal-to-noise ratio, and noise values are the parameters corresponding to the detected target; the presence degree is 1; It is set to the maximum value for the speed direction, otherwise it is 0; the number of consecutive detections is set to 1.
  • the first identification module 611 further includes a judgment submodule 6115 and an update submodule 6116.
  • the judging sub-module 6115 is configured to judge whether the time point corresponding to the current frame is the initialization end time point of the millimeter wave radar.
  • the update submodule 6116 is configured to update the parameters of all the targets measured in the reference object map if the time point corresponding to the current frame is the end time point of the millimeter wave radar initialization.
  • the specific method for updating the parameters of all the measured targets in the reference object map includes: decreasing the existence heat of each measured target in the reference object map within a preset range, and the preset range is 0 to The preset value, the decrement process is also decremented according to the preset gradient, until it is reduced to 0, that is, the measured target's existence heat is updated once after each frame of data is processed.
  • the existing target in the reference map does not have a corresponding detection target in the frame, when the number of consecutive detections is positive, set it to -1; when the number of consecutive detections is negative, reduce the value by 1.
  • the updated sub-module 6116 is also configured to mark the measured target whose presence heat is higher than the preset value and the recent movement heat is 0 as a stationary object; and mark the measured target whose number of consecutive detections reaches the preset minimum value as non-stationary Object; and delete the measured target with a heat degree of 0 from the reference object map.
  • the second identification module 612 includes: an acquisition submodule 6121, a determination submodule 6122, a judgment submodule 6123, a merge submodule 6124, and a creation submodule 6125.
  • the obtaining submodule 6121 is configured to obtain a detection target of one frame of detection information in the multi-frame detection information.
  • the determination submodule 6122 is configured to determine the detection target as a human target if the detection target obtained by the acquisition submodule 6121 is marked as a non-stationary object in the reference object map.
  • the determining sub-module 6123 is configured to determine whether the plane projection distance between the human body target determined by the determining sub-module 6122 and the existing target in the human body map is less than a threshold.
  • the merging submodule 6124 is configured to, if the judging submodule 6123 determines that it is less than the threshold, merge the human body target and the existing target into a new human body target, and update the parameters of the new human body target in the human body map.
  • the creation sub-module 6125 is configured to, if the judging sub-module 6123 determines that it is not less than the threshold, create a new human target corresponding to the human target in the human map, and determine the parameters of the new human target according to the parameters of the human target.
  • the second recognition module 612 further includes: a comparison sub-module 6126 configured to determine a human body target determined from one frame of detection information and multiple frames of detection information before the one frame of detection information. comparing.
  • the merging sub-module 6124 is further configured to, if the comparison sub-module 6126 determines that the planar projection distance of the human target in the human body map is less than the threshold, merge the human target determined by the multi-frame detection information, and update the merge The parameters of the posterior human body target in the human body map.
  • the determining submodule 6122 is further configured to determine the height of the human body target by using the spatial coordinates of the human body target and the body map.
  • the determining sub-module 6122 is also used to, if the detection target is marked as a non-stationary object in the reference object map, and the detection target has other detections that are also marked as a non-stationary object in a specified range in the human body map Target, the detection target and other detection targets are jointly determined as human targets.
  • Some embodiments of the present disclosure provide a device for human body gesture recognition, which can quickly recognize gesture actions made by the human body based on the human body and its hands simultaneously recognized by the millimeter wave radar.
  • This embodiment corresponds to the foregoing method embodiment.
  • this embodiment will not repeat the details in the foregoing method embodiment one by one, but it should be clear that the device in this embodiment can correspondingly implement the foregoing method embodiment.
  • the device 7 includes a first determining unit 71, a second determining unit 72, a fitting unit 73, and a gesture recognition unit 74.
  • the first determining unit 71 is configured to use the hand target corresponding to the human target detected by the millimeter wave radar to determine the spatial position and radial velocity of the hand target in a plurality of consecutive frames.
  • the second determining unit 72 is configured to determine the movement trajectory of the hand target according to the spatial position and the radial velocity determined by the first determining unit 71.
  • the fitting unit 73 is configured to perform linear fitting on the movement trajectory when the movement trajectory determined by the second determination unit 72 reaches a preset length.
  • the gesture recognition unit 74 is configured to determine the human body gesture of the human target according to the direction vector and the radial velocity of the straight line fitted by the fitting unit 73.
  • the gesture recognition unit 74 is further configured to, after determining a human gesture of the human target, pause the recognition of the human gesture corresponding to the human target for a preset period of time.
  • the fitting unit 73 is configured to perform a spatial straight line fitting on the movement trajectory when the movement trajectory determined by the second determination unit 72 reaches a preset length.
  • the fitting unit 73 is configured to project the movement trajectory onto a preset plane when the movement trajectory determined by the second determining unit 72 reaches a preset length, and perform straight-line fitting in the preset plane.
  • the device further includes a setting unit 75 configured to set a preset length of the motion track, where the preset length is the number of hand targets in different spatial positions.
  • the first determining unit 71 is further configured to obtain the spatial position and radial velocity of the hand target in multiple consecutive frames according to the preset length set by the setting unit 75.
  • the device further includes an update unit 76 configured to update multiple consecutive frames corresponding to the motion trajectory when the gesture recognition unit 74 cannot determine the human body gesture of the human target according to the direction vector and the radial velocity.
  • the second determining unit 72 includes: a determining module 721 and a determining module.
  • the determining module 721 is configured to determine whether the number of consecutive frames in which the hand target is not detected is greater than the preset number of frames when the hand target is not continuously detected in a plurality of consecutive frames.
  • the determining module 722 is configured to stop determining the movement trajectory of the hand target when the determining module 722 determines that the number of frames is greater than the preset number of frames.
  • Some embodiments of the present disclosure further provide a processor configured to run a program, wherein the program executes one or more of the human hand recognition methods described in any of the above embodiments when the program is running Step, or execute one or more steps in the method for human gesture recognition described in any of the above embodiments.
  • the processor may be a central processing unit (Central Processing Unit, CPU for short), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASICs), field programmable gate arrays (FPGAs). ) Or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc.
  • the general-purpose processor may be a microprocessor or the processor may also be any conventional processor.
  • Some embodiments of the present disclosure provide a display device 8, as shown in FIG. 13, including a display panel 81; a millimeter-wave radar 82 integrated in the display panel 81, and the millimeter-wave radar 82 is configured to Collecting multiple frames of detection information within a time period; and the processor 83 described in any of the foregoing embodiments.
  • the millimeter-wave radar appears as a flat-shaped object, which actually includes a complete system of transceiver antennas, radio frequency front-ends, data processors, power supplies, output interfaces, etc., which are composed of discrete or integrated chips.
  • the transceiving antennas are arranged in a two-dimensional array on the flat panel.
  • the transmitting and receiving antennas are placed vertically, and the normal direction is in the horizontal plane, so that a more reasonable detection range can be obtained and calculations are more convenient.
  • the millimeter-wave radar is a frequency modulated continuous wave (FMCW) millimeter-wave radar, which may be the IWR6843 radar chip and supporting antenna of Texas Instruments (TI), and the working frequency range is 60GHz ⁇ 64GHz, with 3 Two transmitting antennas and four receiving antennas can output the measured three-dimensional coordinates and radial velocity of the target, and also include a data processor that can perform data calculation and processing.
  • FMCW frequency modulated continuous wave
  • TI Texas Instruments
  • Some embodiments of the present disclosure provide a computer program.
  • the computer program instructions When the computer program executes the computer program instructions on a computer, the computer program instructions cause the computer to execute the method for human hand recognition as described in any of the above embodiments.
  • Some embodiments of the present disclosure also provide a computer program product.
  • the computer program product includes computer program instructions.
  • the computer program instructions When the computer program instructions are executed on a computer, the computer program instructions cause the computer to execute the computer program as described in any of the above-mentioned embodiments.
  • Some embodiments of the present disclosure also provide a computer-readable storage medium, the computer-readable storage medium stores computer program instructions, and when the computer program instructions run on a processor, the processor executes as described in any of the above-mentioned embodiments.
  • One or more steps in the method for human hand recognition, or one or more steps in the method for human gesture recognition in any one of the above embodiments are performed.
  • computer-readable storage media includes permanent and non-permanent, removable and non-removable media, and information storage can be realized by any method or technology.
  • the information can be computer-readable instructions, data structures, program modules, or other data.
  • Exemplary, examples of computer-readable storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), and other types of random access memory (RAM). ), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) Or other optical storage, magnetic cassette tape, disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by computing devices. As defined in this article, computer-readable storage media does not include transitory media, such as modulated data signals and carrier waves.
  • PRAM phase change memory
  • SRAM static random access memory
  • DRAM dynamic random access memory
  • RAM random access memory
  • ROM read-only memory
  • EEPROM electrically erasable programmable read-only memory
  • flash memory or other memory technologies
  • CD-ROM compact disc
  • DVD digital versatile disc
  • computer-readable storage media does not include transitory media
  • the embodiments of the present disclosure can be provided as a method, a system, or a computer program product. Therefore, the present disclosure may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present disclosure may take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program codes.
  • a computer-readable storage media including but not limited to disk storage, CD-ROM, optical storage, etc.
  • These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing equipment to work in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including the instruction device.
  • the device implements the functions specified in one process or multiple processes in the flowchart and/or one block or multiple blocks in the block diagram.
  • These computer program instructions can also be loaded on a computer or other programmable data processing equipment, so that a series of operation steps are executed on the computer or other programmable equipment to produce computer-implemented processing, so as to execute on the computer or other programmable equipment.
  • the instructions provide steps for implementing functions specified in a flow or multiple flows in the flowchart and/or a block or multiple blocks in the block diagram.
  • the computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
  • processors CPUs
  • input/output interfaces network interfaces
  • memory volatile and non-volatile memory
  • the memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and/or non-volatile memory, such as read-only memory (ROM) or flash memory (flash RAM).
  • RAM random access memory
  • ROM read-only memory
  • flash RAM flash memory
  • the embodiments of the present disclosure can be provided as a method, a system, or a computer program product. Therefore, the present disclosure may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present disclosure may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.
  • computer-usable storage media including but not limited to disk storage, CD-ROM, optical storage, etc.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Remote Sensing (AREA)
  • General Physics & Mathematics (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • General Engineering & Computer Science (AREA)
  • Human Computer Interaction (AREA)
  • Health & Medical Sciences (AREA)
  • Electromagnetism (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Health & Medical Sciences (AREA)
  • Psychiatry (AREA)
  • Social Psychology (AREA)
  • Multimedia (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

一种人体手部识别的方法，包括：利用毫米波雷达在预设时间段内采集的多帧检测信息识别人体目标；根据当前帧的检测信息，以人体目标为中心确定在预设范围内是否存在满足设定条件的新检测目标，设定条件包括：具有径向速度；若存在，则将满足设定条件的新检测目标确定为人体目标的手部；若不存在，则确定当前帧中不存在人体目标的手部。

Description

人体手部与手势识别的方法及装置、显示设备
本申请要求于2020年05月26日提交中国专利局、申请号为202010455620.X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
本公开涉及雷达检测技术领域，例如涉及一种人体手部与手势识别的方法及装置、显示设备。
背景技术
毫米波雷达，是指工作频段在毫米波频段的雷达。由于毫米波的波长介于厘米波和光波之间，因此毫米波兼有微波制导和光电制导的优点。与红外、激光、电视等光学导引头相比，毫米波导引头穿透雾、烟、灰尘的能力强，具有全天候全天时的特点。毫米波雷达是测量被测物体相对距离、相对速度、方位的高精度传感器，早期被应用于军事领域，随着雷达技术的发展与进步，毫米波雷达传感器开始应用于汽车电子、无人机、智能交通等多个领域。
发明内容
一方面，提供一种人体手部识别的方法。所述方法包括：利用毫米波雷达在预设时间段内采集的多帧检测信息识别人体目标；根据当前帧的检测信息，以所述人体目标为中心确定在预设范围内是否存在满足设定条件的新检测目标，所述设定条件包括：具有径向速度；若存在，则将满足设定条件的新检测目标确定为所述人体目标的手部；若不存在，则确定所述当前帧中不存在所述人体目标的手部。
在一些实施例中，所述利用毫米波雷达在预设时间段内采集的多帧检测信息识别人体目标，包括：基于所述多帧检测信息中的检测目标生成和维护参考物体图，利用所述参考物体图识别所述毫米波雷达检测到的目标为静止物体或非静止物体；基于所述参考物体图识别的非静止物体生成和维护人体地图，利用所述人体地图从所述多帧检测信息中的非静止物体中确定人体目标。
在一些实施例中，所述基于所述多帧检测信息中的检测目标生成和维护参考物体图，包括：基于所述多帧检测信息中的检测目标的参数生成和维护参考物体图，所述参数包括：空间坐标、径向速度、信噪比、噪声、存在热度、最近运动热度以及连续检测次数中的至少一种。
在一些实施例中，所述存在热度，用于表示检测目标在多帧检测信 息中被检测到的频次，频次越高则存在热度越大；和/或，所述最近运动热度，用于表示检测目标被检测到运动的最后一帧与当前帧之间的时间间隔，时间间隔越小则最近运动热度的值越大；和/或，所述连续检测次数，用于表示检测目标被连续检测到或连续未检测到的次数；所述连续检测次数为正值，表示被连续检测到，所述连续检测次数为负值，表示连续未检测到。
在一些实施例中，所述维护参考物体图，包括：按照检测顺序逐一提取所述多帧检测信息的每一帧中的检测目标；根据所述检测目标的空间位置对一帧检测信息中的检测目标逐一与参考物体图中的已有目标计算空间距离；若所述空间距离小于第一阈值，则确定所述检测目标与空间距离最小的已有目标为同一测得目标；若所述空间距离不小于第一阈值，则在所述参考物体图中为所述检测目标创建测得目标。
在一些实施例中，在确定所述检测目标与空间距离最小的已有目标为同一测得目标之后，还包括：更新所述已有目标的参数：空间坐标更新为检测目标与已有目标的空间坐标的平均值；径向速度、信噪比与噪声更新为检测目标对应的参数值；存在热度在预设范围内基于帧数对应的梯度值升高；最近运动热度在具有径向速度时更新为最近运动热度的最大值；连续检测次数在预设范围内基于上一帧的检测状态更新；若连续检测次数达到最大值，则将存在热度设为最大值，最近运动热度设为0。
在一些实施例中，在更新所述已有目标的参数之后，还包括：若所述已有目标的存在热度高于预设值且最近运动热度为0，则将其标记为静止物体；和/或，若所述已有目标的连续检测次数达到预设最小值，则将其标记为非静止物体。
在一些实施例中，在所述参考物体图中为所述检测目标创建测得目标之后，还包括：确定所创建的测得目标的参数：其中，空间坐标、径向速度、信噪比与噪声的值为检测目标对应的参数值；存在热度为1；最近运动热度在具有径向速度时设置为最大值，否则为0；连续检测次数设置为1。
在一些实施例中，所述维护参考物体图，还包括：若当前帧对应的时间点为所述毫米波雷达的初始化过程的结束时间点，则更新所述参考物体图中所有测得目标的参数：将各测得目标的存在热度在预设范围内递减；若所述参考物体图中的已有目标在当前帧中不存在对应的测得目 标，则在所述已有目标的连续检测次数为正值时，将其设置为-1，在所述已有目标的连续检测次数为负值时，将数值减1；将所述存在热度高于预设值且最近运动热度为0的测得目标标记为静止物体；将所述连续检测次数达到预设最小值的测得目标标记为非静止物体；将所述存在热度为0的测得目标从所述参考物体图中删除。所述初始化过程，为毫米波雷达启动从第一个检测帧出现时开始，到达预设时间时结束，处理采集的多帧检测信息，以生成和维护参考物体图的过程。
在一些实施例中，所述维护人体地图，包括：获取所述多帧检测信息中的一帧检测信息的检测目标；若所述检测目标在所述参考物体图中被标记为非静止物体，则将所述检测目标确定为人体目标。
在一些实施例中，所述维护人体地图，包括：获取所述多帧检测信息中的一帧检测信息的检测目标；若所述检测目标在所述参考物体图中被标记为非静止物体，且所述检测目标在所述人体地图中指定的范围内存在同样被标记为非静止物体的其他检测目标，则将所述检测目标与其他检测目标共同确定为人体目标。
在一些实施例中，在确定所述人体目标之后，还包括：判断所述人体目标与所述人体地图中的已有目标的平面投影距离是否小于第二阈值；若小于，则将所述人体目标与所述已有目标合并为新人体目标，并更新所述人体地图中所述新人体目标的参数；若不小于，则在所述人体地图中创建所述人体目标对应的新人体目标，并根据所述人体目标的参数确定所述新人体目标的参数。
在一些实施例中，所述维护人体地图，还包括：将从所述一帧检测信息中确定的人体目标与所述一帧检测信息之前的多帧检测信息共同确定的人体目标进行对比；若所对比的人体目标在人体地图中的平面投影距离小于第三阈值，则将多帧检测信息所确定的人体目标进行合并，并更新合并后人体目标在人体地图中的参数。
在一些实施例中，所述将满足设定条件的新检测目标确定为所述人体目标的手部，还包括：若满足设定条件的新检测目标为多个时，则根据毫米波雷达与满足设定条件的各新检测目标的距离确定所述人体目标的手部对应的新检测目标。
在一些实施例中，所述设定条件，还包括：信噪比的回波信号强度低于人体目标。
另一方面，提供一种人体手势识别的方法。所述人体手势识别的方 法包括：利用毫米波雷达检测到的人体目标对应的手部目标确定在多个连续帧中所述手部目标的空间位置与径向速度；根据所述空间位置与径向速度确定所述手部目标的运动轨迹；在所述运动轨迹达到预设长度时，对所述运动轨迹进行直线拟合；根据拟合得到的直线的方向矢量以及径向速度确定所述人体目标的人体手势。
在一些实施例中，所述直线拟合，包括：空间直线拟合；和/或，将所述运动轨迹投影至预设平面中，在所述预设平面内进行直线拟合。
在一些实施例中，所述根据所述空间位置与径向速度确定所述手部目标的运动轨迹，包括：在所述手部目标在所述多个连续帧中未被检测到时，判断未检测到所述手部目标的连续帧数是否大于预设帧数；若未被检测到所述手部目标的连续帧数大于预设帧数，则停止确定所述手部目标的运动轨迹。
再一方面，提供一种人体手部识别的装置。所述装置包括：识别单元、判断单元和确定单元。识别单元被配置为利用毫米波雷达在预设时间段内采集的多帧检测信息识别人体目标。判断单元被配置为根据当前帧的检测信息，以所述识别单元确定的人体目标为中心确定在预设范围内是否存在满足设定条件的新检测目标；所述设定条件包括：具有径向速度。确定单元被配置为若所述判断单元确定存在满足设定条件的新检测目标，则将其确定为所述人体目标的手部；以及还被配置为若所述判断单元确定不存在满足设定条件的新检测目标，则确定所述当前帧中不存在所述人体目标的手部。
又一方面，提供一种人体手势识别的装置。包括第一确定单元、第二确定单元、拟合单元和手势识别单元。第一确定单元被配置为利用毫米波雷达检测到的人体目标对应的手部目标确定在多个连续帧中所述手部目标的空间位置与径向速度。第二确定单元被配置为根据所述第一确定单元确定的空间位置与径向速度确定所述手部目标的运动轨迹。拟合单元被配置为在所述第二确定单元确定的运动轨迹达到预设长度时，对所述运动轨迹进行空间直线拟合。手势识别单元被配置为根据所述拟合单元拟合得到的空间直线的方向矢量以及径向速度确定所述人体目标的人体手势。
又一方面，提供一种处理器，用于运行程序，所述程序运行时执行上述任一实施例中所述的人体手部识别的方法，或者是执行上述任一实施例中所述的人体手势识别的方法。
又一方面，提供一种显示设备。所述显示设备包括：显示面板；集成在所述显示面板中的毫米波雷达，所述毫米波雷达被配置为在预设时间段内采集多帧检测信息；以及如上述任一实施例中所述的处理器。
再一方面，提供一种计算机可读存储介质。所述计算机可读存储介质存储有计算机程序指令，所述计算机程序指令在处理器上运行时，使得所述处理器执行如上述任一实施例中所述的人体手部识别的方法中的一个或多个步骤；或者是执行上述任一实施例中所述的人体手势识别的方法中的一个或多个步骤。
又一方面，提供一种计算机程序产品。所述计算机程序产品包括计算机程序指令，在计算机上执行所述计算机程序指令时，所述计算机程序指令使计算机执行如上述任一实施例中所述的人体手部识别的方法中的一个或多个步骤；或者是执行如上述任一实施例中所述的人体手势识别的方法中的一个或多个步骤。
又一方面，提供一种计算机程序。当所述计算机程序在计算机上执行时，所述计算机程序使计算机执行如上述任一实施例中所述的人体手势识别的方法中的一个或多个步骤；或者是执行如上述任一实施例中所述的人体手势识别的方法中的一个或多个步骤。
附图说明
为了更清楚地说明本公开中的技术方案，下面将对本公开一些实施例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的附图，对于本领域普通技术人员来讲，还可以根据这些附图获得其他的附图。此外，以下描述中的附图可以视作示意图，并非对本公开实施例所涉及的产品的实际尺寸、方法的实际流程、信号的实际时序等的限制。
图1为根据一些实施例的一种人体手部识别的方法的流程图；
图2为根据一些实施例的一种利用毫米波雷达在预设时间段内采集的多帧检测信息识别人体目标的方法的流程图；
图3为根据一些实施例的一种维护参考物体图的方法的流程图；
图4为根据一些实施例的另一种维护参考物体图的方法的流程图。
图5为根据一些实施例的一种维护人体地图的方法的流程图；
图6为根据一些实施例的另一种维护人体地图的方法的流程图；
图7为根据一些实施例的一种人体手势识别的方法的流程图；
图8为根据一些实施例的一种根据所述空间位置与径向速度确定所述 手部目标的运动轨迹的方法的流程图；
图9为根据一些实施例的一种人体手部识别的装置的结构图；
图10为根据一些实施例的另一种人体手部识别的装置的结构图；
图11为根据一些实施例的一种人体手势识别的装置的结构图；
图12为根据一些实施例的另一种人体手势识别的装置的结构图；
图13为根据一些实施例的一种显示设备的结构图。
具体实施方式
下面将结合附图，对本公开一些实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本公开一部分实施例，而不是全部的实施例。基于本公开所提供的实施例，本领域普通技术人员所获得的所有其他实施例，都属于本公开保护的范围。
除非上下文另有要求，否则，在整个说明书和权利要求书中，术语“包括(comprise)”及其其他形式例如第三人称单数形式“包括(comprises)”和现在分词形式“包括(comprising)”被解释为开放、包含的意思，即为“包含，但不限于”。在说明书的描述中，术语“一个实施例(one embodiment)”、“一些实施例(some embodiments)”、“示例性实施例(exemplary embodiments)”、“示例(example)”、“特定示例(specific example)”或“一些示例(some examples)”等旨在表明与该实施例或示例相关的特定特征、结构、材料或特性包括在本公开的至少一个实施例或示例中。上述术语的示意性表示不一定是指同一实施例或示例。此外，所述的特定特征、结构、材料或特点可以以任何适当方式包括在任何一个或多个实施例或示例中。
以下，术语“第一”、“第二”仅用于描述目的，而不能理解为指示或暗示相对重要性或者隐含指明所指示的技术特征的数量。由此，限定有“第一”、“第二”的特征可以明示或者隐含地包括一个或者更多个该特征。在本公开实施例的描述中，除非另有说明，“多个”的含义是两个或两个以上。
“A、B和C中的至少一个”与“A、B或C中的至少一个”具有相同含义，均包括以下A、B和C的组合：仅A，仅B，仅C，A和B的组合，A和C的组合，B和C的组合，及A、B和C的组合。
“A和/或B”，包括以下三种组合：仅A，仅B，及A和B的组合。
如本文中所使用，根据上下文，术语“如果”任选地被解释为意思是“当……时”或“在……时”或“响应于确定”或“响应于检测到”。类似地，根据上下文，短语“如果确定……”或“如果检测到[所陈述的条件或事件]”任选地被解释为是指“在确定……时”或“响应于确定……”或“在检测到[所 陈述的条件或事件]时”或“响应于检测到[所陈述的条件或事件]”。
本文中“适用于”或“被配置为”的使用意味着开放和包容性的语言，其不排除适用于或被配置为执行额外任务或步骤的设备。
另外，“基于”的使用意味着开放和包容性，因为“基于”一个或多个所述条件或值的过程、步骤、计算或其他动作在实践中可以基于额外条件或超出所述的值。
毫米波雷达对运动目标的进行检测时具有一定的优势，例如可以用于人体检测，或者是可以用于手部检测等，其中，不论是人体目标或者是手部目标都是因为其与毫米波雷达的相对运动才可以被准确地识别检测到。但是，在人体与其手部同时存在运动时，由于两者的距离相对较近，因而两者的运动会产生相互干扰，使得无法有效分辨出运动中的人体或手部，也就无法识别出手部做出的手势。
基于此，本公开一些实施例提供了一种人体手部识别的方法，该方法可以对人体及其手部同时进行识别检测，如图1所示，该方法包括S1～S4。
S1、利用毫米波雷达在预设时间段内采集的多帧检测信息识别人体目标。
本步骤是基于毫米波雷达的多帧检测信息进行分析，通过对比不同帧中的测得目标来确定人体目标。示例性的，在本步骤中，需要对每个测得目标设置多个参数，其中，包括常规参数与自定义参数，常规参数如空间坐标、径向速度(即测得目标与毫米波雷达之间的空间距离的变化速率)、信噪比与噪声等，而自定义参数主要用于描述一个测得目标在多个连续帧中的运动状态，以此来辅助判断测试目标是静止物体还是非静止物体，并进一步确定是否为人体目标。需要说明的是，这里的非静止物体包括在检测范围内运动的物体以及从检测范围内移出的物体。
S2、根据当前帧的检测信息，以人体目标为中心确定在预设范围内是否存在满足设定条件的新检测目标；设定条件包括：具有径向速度。
本步骤中以当前帧的检测信息作为分析的内容，是因为在多帧检测信息中，同一个人体目标的位置有可能发生变动，导致以该人体目标为中心所确定的预设范围不准确，从而无法有效确定出新检测目标。
需要说明的是，毫米波雷达检测到的测得目标是具有空间坐标的目标，当人的手位于人体与毫米波雷达之间时，虽然手可能会遮挡部分人体，但通过空间坐标的位置分析还是可以将人体目标与手部目标进行区分的。
本步骤中的新检测目标是在当前帧中出现在人体目标附近的可被毫米波雷达检测到的目标。本步骤中的当前帧为S1中多帧检测信息中的一帧检测信 息；本步骤中的径向速度用于表示新检测目标与毫米波雷达之间的空间距离的变化速率。
在一些示例中，设定条件还包括：信噪比低于人体目标的信噪比。
这样，当确定存在新检测目标时，除了根据该新检测目标是否具有径向速度(即是否运动)来识别其是否为该人体目标的手部之外，还结合该新检测目标的信噪比是否低于人体目标来识别，以使判断结果更加准确，这是因为人体手部的等效反射面积小，回波信号强度小；而人体躯干的等效反射面积大，回波信号强度大。在此基础上，在噪声强度接近的情况下，能表现为人体手部的信噪比低于人体目标(即人体躯干)。
当存在满足设定条件的新检测目标时，则该新检测目标可以被确定为该人体目标的候选手部，此时执行S3；反之，当不存在满足设定条件的新检测目标时，则执行S4。
S3、将满足设定条件的新检测目标确定为人体目标的手部。
本步骤中，识别人体目标的手部后，可以进一步识别手势。
需要说明的是，如果该新检测目标不满足上述设定条件，则可确定其不是手部；或者确定其为手但并没有做出待识别的手势，此处对这类新检测目标不做识别。
示例性的，若确定满足设定条件的新检测目标为多个时，则根据毫米波雷达与满足设定条件的各新检测目标的距离确定哪个新检测目标为人体目标的手部。
例如，在毫米波雷达的测量范围内，将距离毫米波雷达最近的满足设定条件的新检测目标确定为人体目标的手部。需要说明的是，将距离毫米波雷达最近的满足设定条件的新检测目标确定为人体目标的手部仅仅是基于上述步骤一般的应用场景作出的选择，因为在识别人体手部时，人体通常是面向毫米波雷达，向前伸出手部并做出动作，在这种情况下，人体目标的手部通常距离毫米波雷达最近。也就是说，若是在其他的应用场景下，通过距离确定哪个新检测目标为人体目标的手部时，也可以采取其他依据，如距离最远等进行判断。
S4、确定当前帧中不存在人体目标的手部。
基于上述的实现方式可以看出，本公开各个实施例所采用的基于毫米波雷达的人体手部识别的方法，是对在预设时间段内由毫米波雷达采集的多帧检测信息进行分析，通过自定义的参数辅助确定测得目标中是否存在人体目标，并针对当前帧的检测信息分析疑似为手部的新检测目标，当确定存在新检测目标时，根据该新检测目标是否满足设定条件来识别其是否为该人体目标 的手部。如此，通过上述步骤可以通过毫米波雷达的检测有效识别出在检测范围内的人体及其手部，以便于后续针对手部的做出进行手势识别。
在一些实施例中，S1中的利用毫米波雷达在预设时间段内采集的多帧检测信息识别人体目标，如图2所示，包括S11～S12。
S11、基于多帧检测信息中的检测目标生成和维护参考物体图，并利用该参考物体图识别毫米波雷达检测到的目标为静止物体或非静止物体。
例如，基于多帧检测信息中的检测目标的参数生成和维护参考物体图。
需要说明的是，本步骤中的参考物体图是基于第一帧检测信息中的检测目标生成的，并且可以基于后续的多帧检测信息中的检测目标进行维护。也即，在处理第一帧检测信息之前，可以不存在参考物体图，根据第一帧检测信息得到的检测目标，生成参考物体图；在处理第二帧检测信息时，第一帧检测信息的检测目标就成为了已有目标，与第二帧检测信息中的检测目标进行距离比对，可以更新参考物体图中的已有目标，以实现参考物体图的维护。以此类推，随着处理帧数的增多，可以不断地向参考物体图中增加新的检测目标。
示例性的，检测目标的参数包括常规参数，如空间坐标、径向速度、信噪比与噪声。
示例性的，检测目标的参数包括自定义参数，如存在热度、最近运动热度以及连续检测次数。其中，存在热度，用于表示检测目标在多帧检测信息中被检测到的频次，频次越高则存在热度越大；最近运动热度，用于表示检测目标被检测到运动的最后一帧与当前帧之间的时间间隔，时间间隔越小则最近运动热度的值越大；连续检测次数，用于表示检测目标被连续检测到或连续未检测到的次数，其中，正值表示被连续检测到，负值表示连续未检测到。
S12、根据参考物体图识别的非静止物体生成和维护人体地图，并利用该人体地图从多帧检测信息中的非静止物体中确定人体目标。
需要说明的是，人体地图与参考物体图类似，是基于第一帧检测信息生成的，并且可以基于后续的多帧检测信息进行维护。也即，在处理第一帧检测信息之前，可以不存在人体地图，根据第一帧检测信息生成人体地图；在处理第二帧检测信息时，第一帧检测信息的非静止物体就成为了已有目标，与第二帧检测信息中的非静止物体目标进行距离比对，可以更新人体地图中的人体目标，以实现人体地图的维护。
人体地图与参考物体图的区别在于参考物体图为三维空间坐标系下的图；而人体地图是一个二维地图，可以为应用场景中的地面，也可以根据需求设置人体地图的所属平面。在人体地图中所记录人体目标的参数包括：平面坐标、 高度(人体身高)、最近运动热度、累积目标数。其中，累积目标数是指雷达所检测到的目标点。
在一些实施例中，S11中的维护参考物体图，如图3所示，包括S111～S116。
S111、按照检测顺序逐一提取多帧检测信息的每一帧中的检测目标。
本步骤是根据采集的帧数逐帧地提取检测目标的数据，包括该检测目标的具体参数；其中，通过获取检测目标的空间坐标以及径向速度的值可以判断该检测目标是否需要进行后续步骤的分析，也就是说判断这些检测目标的合理性，确定该检测目标为真实存在的物体。
示例性的，判断这些检测目标的合理性的依据是检测目标与毫米波雷达之间的距离是否在预设检测范围内；其中，该预设检测范围是根据实际应用需求而设置的。
S112、根据检测目标的空间位置对一帧检测信息中的检测目标逐一与参考物体图中的已有目标计算空间距离。
本步骤将每一帧检测信息中的每个检测目标逐一与参考物体图进行比对，计算其与已有目标计算空间距离，从而确定该检测目标是否为参考物体图中的已有目标。
需要说明的是，毫米波雷达检测到的检测目标在参考物体图中可以表示为一个点，而该点的大小可以基于雷达的分辨率确定，也可以根据该分辨率人为设置。如此，对于一个人体目标，其在参考物体图中的表示将成为多个检测目标的集合。
S113、若空间距离小于第一阈值，则确定检测目标与已有目标为同一测得目标。
本步骤中，由于检测目标和已有目标之间的距离足够接近，因此将这两个检测目标视为同一个测得目标。
S114、更新该已有目标的参数。
示例性的，更新该已有目标的参数的方式，可以包括：空间坐标更新为检测目标与已有目标空间坐标的平均值(即更新物体在图中的坐标位置)；径向速度、信噪比与噪声更新为检测目标对应的参数值；存在热度在预设范围内基于帧数对应的梯度值升高(比如，预设范围0～100，梯度值为1，那么当确定检测目标与已有
目标为同一目标时，该存在热度的值加1，直至增加到最大值时不再增加)；最近运动热度在具有径向速度时更新为最近运动热度的最大值(最近运动热度也是一个范围值，当检测目标在多帧检测信息中存在且运动时，该最近运动 热度的值就越大，而当检测目标停止运动时，则该最近运动热度的值就降低)；连续检测次数在预设范围内基于上一帧的检测状态更新，由于连续检测次数可以表示两种状态(连续检测到与连续未检测到)，因此，其更新是根据上一帧的值(已有目标的值)确定的，当上一帧的值为正(连续检测到)时，该值加1，当上一帧的值为负(连续未检测到)时，将该值设置为1，该连续检测次数也是一个范围值(多帧检测信息的帧数)，一般为1至最大值(连续检测到)或-1至最小值(连续未检测到)。
当连续检测次数为最大值时，将该检测目标的存在热度设为最大值，最近运动热度设为0。
示例性的，若更新后的已有目标的存在热度高于预设值且最近运动热度为0，即说明该已有目标所对应的物体可以被一直检测到，且未出现运动，此时标记该已有目标为静止物体。
需要说明的是，存在热度的预设值可以人为自定义设置。
示例性的，若更新后的已有目标的连续检测次数达到预设最小值，即说明该已有目标曾经出现在参考物体图中，但由于在一段较长的时间内均不能在相应的位置检测到该已有目标，也就是该已有目标出现了移动，此时标记该已有目标为非静止物体。
示例性的，若本步骤中确定出检测目标同时与多个已有目标的空间距离都小于第一阈值时，将其与最近的已有目标进行合并，同时更新参数值；或者，将多个已有目标与该检测目标全部合并为一个目标，再更新该目标的参数。
S115、若空间距离不小于第一阈值，则在参考物体图中为检测目标创建测得目标。
本步骤中，由于检测目标与所有已有目标的空间距离都不小于第一阈值，说明该检测目标是该帧中新出现的目标，此时就在参考物体图中为该检测目标创建一个测得目标。
S116、确定所创建的测得目标的参数。
示例性的，确定该测得目标的参数的方式，可以包括：空间坐标、径向速度、信噪比与噪声的值为检测目标对应的参数值；存在热度设置为1；最近运动热度在具有径向速度时设置为最大值，否则为0；连续检测次数设置为1。
需要说明的是，上述一些实施例中的第一阈值可以是雷达的分辨率，也可以是人工设置的值。本公开对此不做限制。
在一些实施例中，S11中的维护参考物体图，如图4所示，除了包括上述实施例中的S111～S116以外，还包括S117。
S117、若当前帧对应的时间点为毫米波雷达的初始化结束时间点，则更新参考物体图中所有测得目标的参数。
本步骤的初始化过程是毫米波雷达启动从第一个检测帧出现时开始，到达预设时间时结束，处理所采集的多帧检测信息，生成参考物体图的过程。即该初始化过程是由人为设置时间长度的毫米波雷达处理检测信息的过程。初始化过程的目的是为了毫米波雷达开始工作后能够快速判断检测范围内的哪些检测目标是静止物体，因为一个检测目标要确定其为静止物体一般需要对其进行较长时间的检测，而本步骤中的初始化过程能够将该检测时间强制性的缩小，如果可运动的物体在该初始化过程中未运动，那么也将该物体标记为静止物体，直至在后续的检测过程中，该物体运动了才会将其标记为非静止物体。
在S117中，更新参考物体图中所有测得目标的参数的方式，可以包括：
将参考物体图中的每个测得目标的存在热度在预设范围内递减，所述预设范围为0至预设值，该递减过程也是根据预设梯度递减，直至减少至0，也就是在每处理完一帧数据后就对测得目标的存在热度更新一次。
若参考物体图中的已有目标在该帧中不存在对应的检测目标(也就是在该帧中没有与该已有目标相合并的检测目标时)，则在连续检测次数为正值时，将其设置为-1；在连续检测次数为负值时，将数值减1。
将存在热度高于预设值且最近运动热度为0的测得目标标记为静止物体；将连续检测次数达到预设最小值(即连续多帧都未被检测到)的测得目标标记为非静止物体；将存在热度为0的测得目标从参考物体图中删除，即删除曾经出现过或偶尔出现的物体。
通过上述实施例可以分析出：如果一个物体完全不动，而且稳定地被检测到(信噪比不足的目标，有时即便确实存在，也会时常不被雷达检测到)，存在热度很快就能达到其最大值，对此，该物体将被认定是静止物体；而如果一个物体完全不动，但因为信噪比不足或雷达偶尔的漏检，间断地被检测到，在足够长时间后，存在热度也会达到其最大值，也会被认定是静止物体；而对于可能会运动但基本静止的物体，在运动过后，经过足够长时间或连续被检测到一段时间，重新被标记为静止物体；如果一个静止物体离开了检测范围，很快参考物体图中就不将其标记为静止物体，再经过足够长时间，由于存在热度归0，参考物体图将去除此物体的记录；如果一个静止物体被其他非静止物体暂时遮挡，很快参考物体图中就不将其标记为静止物体，但在经过足够长时间之前(即被删除之前)不再被遮挡，又会被重新标记为静止物体。由此，人体只 要在存在热度最大值对应的时间段内有微小的运动，就不会被识别为静止物体，而被标记为非静止物体。
在一些实施例中，S12中的维护人体地图，如图5所示，包括S121～S125。
S121、获取多帧检测信息中一帧检测信息的检测目标。本步骤中针对一帧检测信息中的检测目标进行处理，这是因为如果累计过去多帧检测信息的数据，当人体运动时，在多个帧中记录了不同位置的人体目标，在生成人体地图时都会被当作疑似的人体目标，导致识别不准确。
S122、若检测目标在参考物体图中被标记为非静止物体，则将该检测目标确定为人体目标。
本步骤是将参考物体图中被标记为非静止物体的检测目标，确定为人体目标。
此外，在一些示例中，可以进一步判断出该被标记为非静止物体的检测目标是否为手部，在未被识别为手部时，确定该检测目标为人体目标。
S123、判断人体目标与人体地图中的已有目标的平面投影距离是否小于第二阈值。本步骤就是将该帧中的人体目标投影到人体地图所在的平面中，并判断其与已有目标之间在该平面中的距离，若平面投影距离小于第二阈值，则执行S124；反之，若平面投影距离不小于第二阈值，则执行S125。
需要说明的是，该第二阈值一般是根据一个人在人体地图平面投影所占面积而设置的，比如0.5m，当小于第二阈值时，就认为该人体目标与已有目标是同一个人体的不同部位。人体地图中的已有目标是指在该帧之前所处理的多帧检测信息而累积得到的目标。
S124、将人体目标与已有目标合并为新人体目标，并更新人体地图中该新人体目标的参数。
本步骤是在当前帧中的人体目标与人体地图中的已有目标之间的距离足够近时，即小于第二阈值时，将两者合为一个目标，即新人体目标。
在一些示例中，更新该新人体目标的参数的具体方式，包括：平面坐标更新为加权平均值(即已有目标的平面坐标按累计目标数加权，该人体目标的平面坐标按1加权)；累积目标数加1；高度更新为人体地图中已有目标当前值和该人体目标高度的较大者；最近运动热度更新为人体地图中该人体目标当前值和已有目标的最近运动热度的较大者。
S125、在人体地图中创建人体目标对应的新人体目标，并根据人体目标的参数确定该新人体目标的参数。
本步骤是在当前帧中的人体目标与人体地图中的已有目标之间的距离不 足够近时，即不小于第二阈值时，在人体地图中创建一个新人体目标，并为该新人体目标设置参数。
在一些示例中，为该新人体目标设置参数的具体方式，包括：平面坐标、高度、最近运动热度均为人体目标的参数，累积目标数设置为1。
在一些实施例中，S12中的维护人体地图，如图6所示，在上述实施例的S121～S125的基础上，将S122替换为S122'。
S122'、若检测目标在参考物体图中被标记为非静止物体，且检测目标在人体地图中指定的范围内存在同样被标记为非静止物体的其他检测目标，则将检测目标与其他检测目标共同确定为人体目标，以此减少对人体目标的逐一分析，提高处理效率。
需要说明的是，指定的范围可以为人体在该人体地图中的投影所占的范围。一般地，在确定人体目标时，检测目标落在人体投影所对应的指定范围内的数量有多个。在一些示例中，当在指定的范围内只有一个检测目标时，则不能将其确定为人体目标，或者是当在指定的范围内的检测目标低于某一数量值时，则不能将其确定为人体目标。
在一些实施例中，如图5和图6所示，S12中的维护人体地图，还包括S126。
S126、将从一帧检测信息中确定的人体目标与该一帧检测信息之前的多帧检测信息共同确定的人体目标进行对比，若所对比的人体目标在人体地图中的平面投影距离小于第三阈值，则将多帧检测信息所确定的人体目标进行合并，并更新合并后人体目标在人体地图中的参数。
上述一些实施例是针对一帧检测信息中的检测目标来更新维护人体地图，当遍历完一帧检测信息中的所有检测目标后，再以帧为单位遍历之前多帧检测信息中所确定的人体目标，如果人体目标在人体地图中的平面投影距离小于第三阈值，该第三阈值为自定义距离，即判断不同帧中确定的人体目标是否足够接近，将多帧检测信息(包括当前帧与之前的多帧检测信息)中所确定的足够接近人体目标进行合并，并更新合并后人体目标在人体地图中的参数。
需要说明的是，S126中更新参数的方式与S124中更新参数的方式相同，此处不再赘述。
通过上述实施例的步骤，即基于参考物体图所识别的非静止物体以及多帧检测信息生成并维护的人体地图，可以确定出具体的人体目标，一般人体目标的最近运动热度不为0。
在一些示例中，基于各个人体目标的高度参数结合对应的空间坐标，还可 以确定出毫米波雷达检测到的人体身高，并根据身高可以进一步地区分所检测的人是大人或小孩。
本公开一些实施例提供了一种人体手势识别的方法，该方法是在上述同时识别出人体以及对应的手部后，快速对手部的动作进行识别以确定人体手势的方法，如图7所示，该方法包括S100～S400。
S100、利用毫米波雷达检测到的人体目标对应的手部目标确定在多个连续帧中该手部目标的空间位置与径向速度。
由于本步骤中的手部目标具有对应的人体目标，如此，在识别到多个人体目标以及多个手部目标时，就可以更准确地判断出哪个人做了什么手势，从而反馈更准确的手势响应结果。
需要说明的是，本步骤是以帧为单元，获取多个连续帧中手部目标的空间位置与径向速度。
在一些示例中，S100中的利用毫米波雷达检测到的人体目标对应的手部目标确定在多个连续帧中该手部目标的空间位置与径向速度，包括S101～S102。
S101、设置运动轨迹的预设长度。
需要说明的是，该预设长度为不同空间位置的手部目标的数量，该数量在多个连续帧中都具有手部目标，并且，每个手部目标的空间位置都不同时，与帧数相同。但是在实际应用中，多个连续帧中存在某一帧因为干扰或其他问题没有检测该手部目标的情况，或者是因手部目标运动较慢导致多个连续帧中的手部目标的空间位置相同，此时，就需要采集更多帧中的手部目标的空间位置与径向速度来构成运动轨迹，使其符合预设长度。
此外，该预设长度的设置还与所要识别手势的难度相关，即简单的手势可以通过较短长度的运动轨迹识别，而复杂的手势就需要较长的运动轨迹来识别。
S102、根据预设长度获取多个连续帧中手部目标的空间位置与径向速度。
本步骤是基于预设长度确定多个连续帧的帧数，进而从多个连续帧中获取手部目标的空间位置与径向速度。
S200、根据空间位置与径向速度确定手部目标的运动轨迹。
本步骤中根据手部目标的空间位置可以确定多个空间点，而根据各个空间位置所对应的径向速度可以确定手部目标的运动方向，将这些空间点以运动方向为参考进行连接，得到该手部目标的运动轨迹。
S300、在运动轨迹达到预设长度时，对运动轨迹进行直线拟合。
本步骤中的运动轨迹是基于多个帧的检测信息确定的，该运动轨迹的长 度与帧数正相关，而要判断手部目标的手势，并不需要获取完整的运动轨迹，可以通过一段运动轨迹进行判断，因此，该预设长度是指通过指定数量的连续帧检测信息所确定的手部目标的运动轨迹。
在一些示例中，在运动轨迹达到预设长度时，对运动轨迹进行空间直线拟合。
示例性的，可以采用最小二乘法对运动轨迹进行空间直线拟合。
在另一些示例中，在运动轨迹达到预设长度时，将运动轨迹投影至预设平面中，在预设平面内进行直线拟合。
需要说明的是，在进行空间直线拟合后，一般会存在两种结果：一种是运动轨迹与空间直线拟合的线性相关度高，即可以拟合为一条空间直线，此时就可以根据所拟合的空间直线进一步判断具体的人体手势。而另一种则是运动轨迹与空间直线拟合的线性相关度不高，即无法拟合为一条空间直线，此时，则将运动轨迹投影至预设平面中，在该预设平面内进行直线拟合；其中，该预设平面是根据实际需求预先设置的，比如毫米波雷达的收发天线所在的平面。
若在预设平面中也无法拟合为直线，则不对该运动轨迹进行人体手势识别。
S400、根据拟合得到的直线的方向矢量以及径向速度确定人体目标的人体手势。
本步骤中所识别的人体手势包括但不限定是人体面向毫米波雷达所做的左右、上下、前后滑动手势。其中，左右、上下滑动手势可以通过空间直线的方向矢量进行识别，而前后滑动手势可以通过手部目标的径向速度进行识别。而将直线的方向矢量以及径向速度相结合则可以识别更多的滑动手势。
在一些示例中，若根据拟合得到的直线的方向矢量以及径向速度不能确定人体目标的人体手势时，更新构成该运动轨迹所对应的多个连续帧，也就是对构成运动轨迹的数据点进行更新。
示例性的，更新构成该运动轨迹所对应的多个连续帧，包括：
去除该运动轨迹对应的第一帧检测信息，再次进行直线拟合；
根据拟合得到的直线的方向矢量以及径向速度确定所述人体目标的人体手势；
重复上述步骤直至所述运动轨迹不能达到预设长度。
本步骤在确保运动轨迹长度满足预设长度的前提下，通过缩短运动轨迹来确定具体的人体手势，如此可以去除手部在做出手势动作之前的额外非手势动作。
在一些示例中，在确定人体目标的一个人体手势后，在预设时间段内暂停对该人体目标识别人体手势。
本步骤是为了避免人体在做出一次手势后，手移动向一下个点准备做第二个手势的过程中被识别为新的手势。
上述实施例是基于对人体以及手部进行同时识别从而得到人体目标对应的手部目标的前提下，实现的对手部目标运动轨迹的追踪，从而快速识别出人体所做出的具体手势。
在一些实施例中，如图8所示，S200中的根据所述空间位置与径向速度确定所述手部目标的运动轨迹，包括S201和S202。
S201、在手部目标在所述多个连续帧中未被检测到时，判断未检测到所述手部目标的连续帧数是否大于帧数。
S202、若未检测到手部目标的连续帧数大于预设帧数，则停止确定手部目标的运动轨迹。
也就是说，上述步骤允许在多个连续帧出现个别帧中未检测到手部目标的情况，但是，对于未检测到手部目标的连续帧数需要作出限定，因为，当连续帧数大于预设帧数时，可以认为是手部目标已经结束该手势，或者是中断了该手势，此时应停止确定该手部目标的运动轨迹。
本公开一些实施例提供了一种人体手部识别的装置，该装置可以利用毫米波雷达同时且准确地识别出被测范围内的人体及其手部目标。本实施例与前述方法实施例对应，为便于阅读，本实施例不再对前述方法实施例中的细节内容进行逐一赘述，但应当明确，本实施例中的装置能够对应实现前述方法实施例中的全部内容。如图9所示，该装置6包括：识别单元61、判断单元62和确定单元63。
识别单元61，被配置为利用毫米波雷达在预设时间段内采集的多帧检测信息识别人体目标。
判断单元62，被配置为根据当前帧的检测信息，以所述识别单元61确定的人体目标为中心确定在预设范围内是否存在满足设定条件的新检测目标，设定条件包括：具有径向速度。
确定单元63，被配置为若所述判断单元62确定存在满足设定条件的新检测目标，则将满足设定条件的新检测目标确定为所述人体目标的手部；以及还被配置为若所述判断单元62确定不存在满足设定条件的新检测目标，则确定当前帧中不存在所述人体目标的手部。
在一些实施例中，上述设定条件还包括：信噪比低于人体目标的信噪比。
在一些实施例中，确定单元63还被配置为，在确定满足设定条件的新检测目标为多个时，根据毫米波雷达与满足设定条件的各新检测目标的距离确定人体目标的手部对应的新检测目标。
示例性的，确定单元63还被配置为，在毫米波雷达的测量范围内，将距离毫米波雷达最近的满足设定条件的新检测目标确定为人体目标的手部。
在一些实施例中，如图10所示，识别单元61包括第一识别模块611和第二识别模块612。
第一识别模块611，被配置为基于多帧检测信息中的检测目标生成和维护参考物体图，利用参考物体图识别毫米波雷达检测到的目标为静止物体或非静止物体。
第二识别模块612，被配置为根据第一识别模块611得到的参考物体图识别的非静止物体生成和维护人体地图，利用人体地图从多帧检测信息中的非静止物体中确定人体目标。
在一些示例中，第一识模块611，被配置为基于多帧检测信息中的检测目标的参数生成和维护参考物体图，利用参考物体图识别毫米波雷达检测到的目标为静止物体或非静止物体。
示例性的，检测目标的参数包括：空间坐标、径向速度、信噪比、噪声、存在热度、最近运动热度以及连续检测次数中的至少一种。
存在热度，用于表示检测目标在多帧检测信息中被检测到的频次，频次越高则存在热度越大。
最近运动热度，用于表示检测目标被检测到运动的最后一帧与当前帧之间的时间间隔，时间间隔越小则最近运动热度的值越大。
连续检测次数，用于表示检测目标被连续检测到或连续未检测到的次数，其中，正值表示被连续检测到，负值表示连续未检测到。
继续参阅图10，第一识别模块611包括提取子模块6111、计算子模块6112、合并子模块6113和创建子模块6114。
提取子模块6111，被配置为按照检测顺序逐一提取多帧检测信息中的每一帧中的检测目标。
计算子模块6112，被配置为根据提取子模块6111得到的检测目标的空间位置对一帧检测信息中的检测目标逐一与参考物体图中的已有目标计算空间距离。
合并子模块6113，被配置为若计算子模块6112得到的空间距离小于阈值，则确定检测目标与空间距离最小的已有目标为同一测得目标；以及还被配 置为在确定检测目标与已有目标为同一测得目标之后，更新已有目标的参数。
创建子模块6114，被配置为若计算子模块6112得到的空间距离不小于阈值，则在参考物体图中为所述检测目标创建测得目标；以及还被配置为在参考物体图中为检测目标创建测得目标之后，确定所创建测得目标的参数。
示例性的，更新已有目标的参数的具体方式，包括：空间坐标更新为检测目标与已有目标空间坐标的平均值；径向速度、信噪比与噪声更新为检测目标对应的参数值；存在热度在预设范围内基于帧数对应的梯度值升高；最近运动热度在具有径向速度时更新为最近运动热度的最大值；连续检测次数在预设范围内基于上一帧的检测状态更新；其中，若连续检测次数达到最大值，则将存在热度设为最大值，最近运动热度设为0。
若存在热度高于预设值且最近运动热度0，则标记已有目标为静止物体；若已有目标的连续检测次数达到预设最小值，则将其标记为非静止物体。
示例性的，确定所创建测得目标的参数的具体方式，包括：空间坐标、径向速度、信噪比与噪声的值为检测目标对应的参数；存在热度为1；最近运动热度在具有径向速度时设置为最大值，否则为0；连续检测次数设置为1。
继续参阅图10，所述第一识别模块611还包括判断子模块6115和更新子模块6116。
判断子模块6115，被配置为判断当前帧对应的时间点是否为毫米波雷达的初始化结束时间点。
更新子模块6116，被配置为若当前帧对应的时间点为毫米波雷达的初始化结束时间点，则更新参考物体图中所有测得目标的参数。
示例性的，更新参考物体图中所有测得目标的参数的具体方式，包括：将参考物体图中的每个测得目标的存在热度在预设范围内递减，所述预设范围为0至预设值，该递减过程也是根据预设梯度递减，直至减少至0，也就是在每处理完一帧数据后就对测得目标的存在热度更新一次。
若参考地图中的已有目标在该帧中不存在对应的检测目标，则在连续检测次数为正值时，将其设置为-1；在连续检测次数为负值时，将数值减1。
更新子模块6116，还被配置为将存在热度高于预设值且最近运动热度为0的测得目标标记为静止物体；以及将连续检测次数达到预设最小值的测得目标标记为非静止物体；以及将存在热度为0的测得目标从参考物体图中删除。
继续参阅图10，第二识别模块612包括：获取子模块6121、确定子模块6122、判断子模块6123、合并子模块6124和创建子模块6125。
获取子模块6121，被配置为获取多帧检测信息中一帧检测信息的检测目 标。
确定子模块6122，被配置为若获取子模块6121得到的检测目标在所述参考物体图中被标记为非静止物体，则将检测目标确定为人体目标。
判断子模块6123，被配置为判断确定子模块6122确定的人体目标与人体地图中的已有目标的平面投影距离是否小于阈值。
合并子模块6124，被配置为若判断子模块6123确定小于阈值，将人体目标与已有目标合并为新人体目标，并更新人体地图中所述新人体目标的参数。
创建子模块6125，被配置为若判断子模块6123确定不小于阈值，在人体地图中创建该人体目标对应的新人体目标，并根据人体目标的参数确定新人体目标的参数。
继续参阅图10，第二识别模块612还包括：对比子模块6126，被配置为将从一帧检测信息中确定的人体目标与所述一帧检测信息之前的多帧检测信息共同确定的人体目标进行对比。
在一些示例中，合并子模块6124还被配置为，若对比子模块6126确定人体目标在人体地图中的平面投影距离小于阈值，则将多帧检测信息所确定的人体目标进行合并，并更新合并后人体目标在人体地图中的参数。
在一些示例中，确定子模块6122还被配置为，利用人体目标的空间坐标以及人体地图确定所述人体目标的身高。
在一些示例中，确定子模块6122还用于，若检测目标在参考物体图中被标记为非静止物体，且检测目标在人体地图中指定的范围内存在同样被标记为非静止物体的其他检测目标，则将检测目标与其他检测目标共同确定为人体目标。
本公开一些实施例提供了一种人体手势识别的装置，该装置可以基于毫米波雷达同时识别出的人体及其手部快速识别出人体所做出的手势动作。本实施例与前述方法实施例对应，为便于阅读，本实施例不再对前述方法实施例中的细节内容进行逐一赘述，但应当明确，本实施例中的装置能够对应实现前述方法实施例中的全部内容。如图11所示，该装置7包括第一确定单元71、第二确定单元72、拟合单元73和手势识别单元74。
第一确定单元71，被配置为利用毫米波雷达检测到的人体目标对应的手部目标确定在多个连续帧中手部目标的空间位置与径向速度。
第二确定单元72，被配置为根据第一确定单元71确定的空间位置与径向速度确定手部目标的运动轨迹。
拟合单元73，被配置为在第二确定单元72确定的运动轨迹达到预设长度 时，对运动轨迹进行直线拟合。
手势识别单元74，用于根据拟合单元73拟合得到的直线的方向矢量以及径向速度确定人体目标的人体手势。
在一些示例中，手势识别单元74还被配置为，在确定人体目标的一个人体手势后，在预设时间段内暂停对人体目标对应的人体手势进行识别。
在一些示例中，拟合单元73，被配置为在第二确定单元72确定的运动轨迹达到预设长度时，对运动轨迹进行空间直线拟合。
或者，拟合单元73，被配置为在第二确定单元72确定的运动轨迹达到预设长度时，将运动轨迹投影至预设平面中，在预设平面内进行直线拟合。
继续参阅图11，所述装置还包括设置单元75，被配置为设置运动轨迹的预设长度，所述预设长度为不同空间位置的手部目标的数量。
在此基础上，第一确定单元71还被配置为，根据设置单元75设置的预设长度获取多个连续帧中手部目标的空间位置与径向速度。
继续参阅图11，所述装置还包括更新单元76，被配置为在手势识别单元74根据方向矢量以及径向速度无法确定人体目标的人体手势时，更新构成运动轨迹所对应的多个连续帧。
继续参阅图11，第二确定单元72包括：判断模块721和确定模块。
判断模块721，被配置为在手部目标在多个连续帧中未被连续检测到时，判断未检测到该手部目标的连续帧数是否大于预设帧数。
确定模块722，被配置为在判断模块722确定大于预设帧数时，停止确定手部目标的运动轨迹。
本公开的一些实施例还提供一种处理器，所述处理器用于运行程序，其中，所述程序运行时执行上述任一实施例中所述的人体手部识别的方法中的一个或多个步骤，或者是执行上述任一实施例中所述的人体手势识别的方法中的一个或多个步骤。
示例性的，处理器可以是中央处理单元(Central Processing Unit，简称CPU)，还可以是其他通用处理器、数字信号处理器(DSP)、专用集成电路(ASIC)、现场可编程门阵列(FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等。其中，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。
本公开的一些实施例提供一种显示设备8，如图13所示，包括显示面板81；集成在所述显示面板81中的毫米波雷达82，所述毫米波雷达82被配置为在预设时间段内采集多帧检测信息；以及上述任一实施例所述的处理器83。
在一些示例中，毫米波雷达表现为一个平板形状物体，实际包含了收发天线、射频前端、数据处理器、电源、输出接口等部分的完整系统，以分立或芯片集成的方式组成。
在一些示例中，收发天线在平板上排布成二维阵列。
在一些示例中，收发天线竖直放置，法向在水平面内，这样可以获得较为合理的检测范围，计算也较便利。
示例性的，毫米波雷达是调频连续波(Frequency Modulated Continuous Wave，FMCW)毫米波雷达，可以是德州仪器(Texas Instruments，TI)公司的IWR6843雷达芯片和配套天线，工作频段60GHz～64GHz，具有3个发射天线和4个接收天线，能够输出测得目标的三维坐标和径向速度，也包含了可以进行数据的运算和处理的数据处理器。
本公开的一些实施例提供一种计算机程序，该计算机程序在计算机上执行该计算机程序指令时，该计算机程序指令使计算机执行如上述任一实施例中所述的人体手部识别的方法中的一个或多个步骤，或者是执行如上述任一实施例中所述的人体手势识别的方法中的一个或多个步骤。本公开的一些实施例还提供一种计算机程序产品，该计算机程序产品包括计算机程序指令，在计算机上执行该计算机程序指令时，该计算机程序指令使计算机执行如上述任一实施例中所述的人体手部识别的方法中的一个或多个步骤，或者是执行如上述任一实施例中所述的人体手势识别的方法中的一个或多个步骤。
本公开的一些实施例还提供一种计算机可读存储介质，该计算机可读存储介质存储有计算机程序指令，计算机程序指令在处理器上运行时，使得处理器执行如上述任一实施例中所述的人体手部识别的方法中的一个或多个步骤，或者是执行上述任一实施例中所述的人体手势识别的方法中的一个或多个步骤。
在一些示例中，计算机可读存储介质包括永久性和非永久性、可移动和非可移动媒体，可以由任何方法或技术来实现信息存储。信息可以是计算机可读指令、数据结构、程序的模块或其他数据。
示例性的，计算机可读存储介质的例子包括，但不限于相变内存(PRAM)、静态随机存取存储器(SRAM)、动态随机存取存储器(DRAM)、其他类型的随机存取存储器(RAM)、只读存储器(ROM)、电可擦除可编程只读存储器(EEPROM)、快闪记忆体或其他内存技术、只读光盘只读存储器(CD-ROM)、数字多功能光盘(DVD)或其他光学存储、磁盒式磁带、磁盘存储或其他磁性存储设备或任何其他非传输介质，可用于存储可以被计算设备访问的信息。按 照本文中的界定，计算机可读存储介质不包括暂存电脑可读媒体(transitory media)，如调制的数据信号和载波。
上述处理器、计算机程序产品、计算机程序以及计算机可读存储介质的有益效果和上述一些实施例所述的人体手部识别的方法、人体手势识别的方法的有益效果相同，此处不再赘述。
在上述各个实施例中，对各个实施例的描述都各有侧重，某个实施例中没有详述的部分，可以参见其他实施例的相关描述。
在此提供的算法和显示不与任何特定计算机、虚拟系统或者其它设备固有相关。各种通用系统也可以与基于在此的示教一起使用。根据上面的描述，构造这类系统所要求的结构是显而易见的。此外，本公开也不针对任何特定编程语言。应当明白，可以利用各种编程语言实现在此描述的本公开的内容，并且上面对特定语言所做的描述是为了披露本公开的最佳实施方式。
本领域内的技术人员应明白，本公开的实施例可提供为方法、系统、或计算机程序产品。因此，本公开可采用完全硬件实施例、完全软件实施例、或结合软件和硬件方面的实施例的形式。而且，本公开可采用在一个或多个其中包含有计算机可用程序代码的计算机可读存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
本公开是参照根据本公开一些实施例的方法、设备(系统)、和计算机程序产品的流程图和/或方框图来描述的。应理解可由计算机程序指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器，使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。
这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理设备以特定方式工作的计算机可读存储器中，使得存储在该计算机可读存储器中的指令产生包括指令装置的制造品，该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。
这些计算机程序指令也可装载到计算机或其他可编程数据处理设备上，使得在计算机或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理，从而在计算机或其他可编程设备上执行的指令提供用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的步骤。
在一个典型的配置中，计算设备包括一个或多个处理器(CPU)、输入/输 出接口、网络接口和内存。
存储器可能包括计算机可读介质中的非永久性存储器，随机存取存储器(RAM)和/或非易失性内存等形式，如只读存储器(ROM)或闪存(flash RAM)。存储器是计算机可读介质的示例。
本领域技术人员应明白，本公开的实施例可提供为方法、系统或计算机程序产品。因此，本公开可采用完全硬件实施例、完全软件实施例或结合软件和硬件方面的实施例的形式。而且，本公开可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
以上所述，仅为本公开的具体实施方式，但本公开的保护范围并不局限于此，任何熟悉本技术领域的技术人员在本公开揭露的技术范围内，想到变化或替换，都应涵盖在本公开的保护范围之内。因此，本公开的保护范围应以所述权利要求的保护范围为准。

Claims (21)

  1. 一种人体手部识别的方法，包括：
    利用毫米波雷达在预设时间段内采集的多帧检测信息识别人体目标；
    根据当前帧的检测信息，以所述人体目标为中心确定在预设范围内是否存在满足设定条件的新检测目标，所述设定条件包括：具有径向速度；
    若存在，则将满足设定条件的新检测目标确定为所述人体目标的手部；
    若不存在，则确定所述当前帧中不存在所述人体目标的手部。
  2. 根据权利要求1所述的方法，其中，所述利用毫米波雷达在预设时间段内采集的多帧检测信息识别人体目标，包括：
    基于所述多帧检测信息中的检测目标生成和维护参考物体图，利用所述参考物体图识别所述毫米波雷达检测到的目标为静止物体或非静止物体；
    基于所述参考物体图识别的非静止物体生成和维护人体地图，利用所述人体地图从所述多帧检测信息中的非静止物体中确定人体目标。
  3. 根据权利要求2所述的方法，其中，所述基于所述多帧检测信息中的检测目标生成和维护参考物体图，包括：
    基于所述多帧检测信息中的检测目标的参数生成和维护参考物体图，所述参数包括：空间坐标、径向速度、信噪比、噪声、存在热度、最近运动热度以及连续检测次数中的至少一种。
  4. 根据权利要求3所述的方法，其中，所述存在热度，用于表示检测目标在多帧检测信息中被检测到的频次，频次越高则存在热度越大；
    和/或，所述最近运动热度，用于表示检测目标被检测到运动的最后一帧与当前帧之间的时间间隔，时间间隔越小则最近运动热度的值越大；
    和/或，所述连续检测次数，用于表示检测目标被连续检测到或连续未检测到的次数；所述连续检测次数为正值，表示被连续检测到，所述连续检测次数为负值，表示连续未检测到。
  5. 根据权利要求3或4所述的方法，其中，所述维护参考物体图，包括：
    按照检测顺序逐一提取所述多帧检测信息的每一帧中的检测目标；
    根据所述检测目标的空间位置对一帧检测信息中的检测目标逐一与参考物体图中的已有目标计算空间距离；
    若所述空间距离小于第一阈值，则确定所述检测目标与空间距离最小的已有目标为同一测得目标；
    若所述空间距离不小于第一阈值，则在所述参考物体图中为所述检测目标创建测得目标。
  6. 根据权利要求5所述的方法，其中，在确定所述检测目标与空间距离最小的已有目标为同一测得目标之后，还包括：
    更新所述已有目标的参数：空间坐标更新为检测目标与已有目标的空间坐标的平均值；径向速度、信噪比与噪声更新为检测目标对应的参数值；存在热度在预设范围内基于帧数对应的梯度值升高；最近运动热度在具有径向速度时更新为最近运动热度的最大值；连续检测次数在预设范围内基于上一帧的检测状态更新；
    若连续检测次数达到最大值，则将存在热度设为最大值，最近运动热度设为0。
  7. 根据权利要求6所述的方法，其中，在更新所述已有目标的参数之后，还包括：
    若所述已有目标的存在热度高于预设值且最近运动热度为0，则将其标记为静止物体；
    和/或，若所述已有目标的连续检测次数达到预设最小值，则将其标记为非静止物体。
  8. 根据权利要求5所述的方法，其中，在所述参考物体图中为所述检测目标创建测得目标之后，还包括：
    确定所创建的测得目标的参数：其中，空间坐标、径向速度、信噪比与噪声的值为检测目标对应的参数值；存在热度为1；最近运动热度在具有径向速度时设置为最大值，否则为0；连续检测次数设置为1。
  9. 根据权利要求5～8中任一项所述的方法，其中，所述维护参考物体图，还包括：
    若当前帧对应的时间点为所述毫米波雷达的初始化过程的结束时间点，则更新所述参考物体图中所有测得目标的参数：将各测得目标的存在热度在预设范围内递减；若所述参考物体图中的已有目标在当前帧中不存在对应的测得目标，则在所述已有目标的连续检测次数为正值时，将其设置为-1，在所述已有目标的连续检测次数为负值时，将数值减1；
    将所述存在热度高于预设值且最近运动热度为0的测得目标标记为静止物体；将所述连续检测次数达到预设最小值的测得目标标记为非静止物体；将所述存在热度为0的测得目标从所述参考物体图中删除；
    所述初始化过程，为毫米波雷达启动从第一个检测帧出现时开始，到达预设时间时结束，处理采集的多帧检测信息，以生成和维护参考物体图的过程。
  10. 根据权利要求2～9中任一项所述的方法，其中，所述维护人体地图，包括：
    获取所述多帧检测信息中的一帧检测信息的检测目标；
    若所述检测目标在所述参考物体图中被标记为非静止物体，则将所述检测目标确定为人体目标。
  11. 根据权利要求2～9任一项所述的方法，其中，所述维护人体地图，包括：
    获取所述多帧检测信息中的一帧检测信息的检测目标；
    若所述检测目标在所述参考物体图中被标记为非静止物体，且所述检测目标在所述人体地图中指定的范围内存在同样被标记为非静止物体的其他检测目标，则将所述检测目标与其他检测目标共同确定为人体目标。
  12. 根据权利要求10或11所述的方法，其中，在确定所述人体目标之后，还包括：
    判断所述人体目标与所述人体地图中的已有目标的平面投影距离是否小于第二阈值；
    若小于，则将所述人体目标与所述已有目标合并为新人体目标，并更新所述人体地图中所述新人体目标的参数；
    若不小于，则在所述人体地图中创建所述人体目标对应的新人体目标，并根据所述人体目标的参数确定所述新人体目标的参数。
  13. 根据权利要求10～12任一项所述的方法，其中，所述维护人体地图，还包括：
    将从所述一帧检测信息中确定的人体目标与所述一帧检测信息之前的多帧检测信息共同确定的人体目标进行对比；
    若所对比的人体目标在人体地图中的平面投影距离小于第三阈值，则将多帧检测信息所确定的人体目标进行合并，并更新合并后人体目标在人体地图中的参数。
  14. 根据权利要求1～13任一项所述的方法，其中，所述将满足设定条件的新检测目标确定为所述人体目标的手部，还包括：
    若满足设定条件的新检测目标为多个时，则根据毫米波雷达与满足设定条件的各新检测目标的距离确定所述人体目标的手部对应的新检测目标。
  15. 根据权利要求1～14任一项所述的方法，其中，所述设定条件，还包括：信噪比低于人体目标的信噪比。
  16. 一种人体手势识别的方法，包括：
    利用毫米波雷达检测到的人体目标对应的手部目标确定在多个连续帧中所述手部目标的空间位置与径向速度；
    根据所述空间位置与径向速度确定所述手部目标的运动轨迹；
    在所述运动轨迹达到预设长度时，对所述运动轨迹进行直线拟合；
    根据拟合得到的直线的方向矢量以及径向速度确定所述人体目标的人体手势。
  17. 根据权利要求16所述的方法，其中，所述直线拟合，包括：
    空间直线拟合；和/或，
    将所述运动轨迹投影至预设平面中，在所述预设平面内进行直线拟合。
  18. 根据权利要求16或17所述的方法，其中，所述根据所述空间位置与径向速度确定所述手部目标的运动轨迹，包括：
    在所述手部目标在所述多个连续帧中未被检测到时，判断未检测到所述手部目标的连续帧数是否大于预设帧数；
    若未检测到所述手部目标的连续帧数大于预设帧数，则停止确定所述手部目标的运动轨迹。
  19. 一种处理器，用于运行程序，所述程序运行时执行权利要求1～15中任一项所述的人体手部识别的方法，或者是执行权利要求16～18中任一项所述的人体手势识别的方法。
  20. 一种显示设备，包括：
    显示面板；
    集成在所述显示面板中的毫米波雷达，所述毫米波雷达被配置为在预设时间段内采集多帧检测信息；以及
    如权利要求19所述的处理器。
  21. 一种计算机可读存储介质，其中，所述计算机可读存储介质存储有计算机程序指令，所述计算机程序指令在处理器上运行时，使得所述处理器执行权利要求1～15中任一项所述的人体手部识别的方法，或者是执行权利要求16～18中任一项所述的人体手势识别的方法。
PCT/CN2021/094191 2020-05-26 2021-05-17 人体手部与手势识别的方法及装置、显示设备 WO2021238710A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US17/790,416 US11797098B2 (en) 2020-05-26 2021-05-17 Methods for recognizing human hand and hand gesture from human, and display apparatus

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010455620.XA CN111624572B (zh) 2020-05-26 2020-05-26 一种人体手部与人体手势识别的方法及装置
CN202010455620.X 2020-05-26

Publications (1)

Publication Number Publication Date
WO2021238710A1 true WO2021238710A1 (zh) 2021-12-02

Family

ID=72258175

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/094191 WO2021238710A1 (zh) 2020-05-26 2021-05-17 人体手部与手势识别的方法及装置、显示设备

Country Status (3)

Country Link
US (1) US11797098B2 (zh)
CN (1) CN111624572B (zh)
WO (1) WO2021238710A1 (zh)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111624572B (zh) * 2020-05-26 2023-07-18 京东方科技集团股份有限公司 一种人体手部与人体手势识别的方法及装置
CN112363156A (zh) * 2020-11-12 2021-02-12 苏州矽典微智能科技有限公司 隔空手势识别方法和装置以及智能设备
CN112788390B (zh) * 2020-12-25 2023-05-23 深圳市优必选科技股份有限公司 一种基于人机交互的控制方法、装置、设备及存储介质
CN113791411B (zh) * 2021-09-07 2024-05-28 北京航空航天大学杭州创新研究院 一种基于轨迹判断的毫米波雷达手势识别方法及装置

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110181509A1 (en) * 2010-01-26 2011-07-28 Nokia Corporation Gesture Control
CN102509088A (zh) * 2011-11-28 2012-06-20 Tcl集团股份有限公司 一种手部运动检测的方法、装置及人机交互系统
CN103699220A (zh) * 2013-12-09 2014-04-02 乐视致新电子科技（天津）有限公司 一种根据手势运动轨迹进行操作的方法及装置
CN105607745A (zh) * 2016-03-16 2016-05-25 京东方科技集团股份有限公司 显示控制电路、显示控制方法和显示装置
CN107961035A (zh) * 2017-12-29 2018-04-27 深圳开立生物医疗科技股份有限公司 一种超声探头以及控制超声诊断仪的方法和装置
CN108200706A (zh) * 2018-01-25 2018-06-22 宁波隔空智能科技有限公司 一种基于微波雷达手势识别技术的照明灯具及其控制方法
CN108344995A (zh) * 2018-01-25 2018-07-31 宁波隔空智能科技有限公司 一种基于微波雷达技术的手势识别装置及手势识别方法
CN109164915A (zh) * 2018-08-17 2019-01-08 湖南时变通讯科技有限公司 一种手势识别方法、装置、系统和设备
US20190049558A1 (en) * 2017-08-08 2019-02-14 KC Sense Limited Hand Gesture Recognition System and Method
CN111624572A (zh) * 2020-05-26 2020-09-04 京东方科技集团股份有限公司 一种人体手部与人体手势识别的方法及装置

Family Cites Families (32)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104714675B (zh) * 2013-12-13 2017-12-05 华为终端（东莞）有限公司 一种手势识别方法及装置
JP6427279B2 (ja) * 2015-04-30 2018-11-21 グーグル エルエルシー ジェスチャの追跡および認識のための、ｒｆに基づいた微細動作追跡
US11592913B2 (en) * 2015-07-17 2023-02-28 Origin Wireless, Inc. Method, apparatus, and system for wireless writing tracking
US10817065B1 (en) * 2015-10-06 2020-10-27 Google Llc Gesture recognition using multiple antenna
CN108604733B (zh) 2016-01-26 2021-07-30 纽威莱克公司 用于手势和动作分析的毫米波传感器系统
DE102016213007A1 (de) * 2016-07-15 2018-01-18 Robert Bosch Gmbh Verfahren und System zur Abtastung eines Objekts
US10579150B2 (en) * 2016-12-05 2020-03-03 Google Llc Concurrent detection of absolute distance and relative movement for sensing action gestures
CN109643372A (zh) * 2017-02-28 2019-04-16 深圳市大疆创新科技有限公司 一种识别方法、设备以及可移动平台
CN107132913A (zh) 2017-03-31 2017-09-05 南京理工大学 一种基于毫米波的多用户人机交互方法
CN107463329B (zh) 2017-07-28 2019-08-27 Oppo广东移动通信有限公司 黑屏手势的检测方法、装置、存储介质及移动终端
CN109697394B (zh) * 2017-10-24 2021-12-28 京东方科技集团股份有限公司 手势检测方法和手势检测设备
CN109712166A (zh) * 2017-10-26 2019-05-03 北京航天长峰科技工业集团有限公司 一种跟踪目标释放系统及方法
CN108509910B (zh) * 2018-04-02 2021-09-28 重庆邮电大学 基于fmcw雷达信号的深度学习手势识别方法
US10775482B2 (en) 2018-04-11 2020-09-15 Infineon Technologies Ag Human detection and identification in a setting using millimeter-wave radar
CN108596092B (zh) * 2018-04-24 2021-05-18 亮风台（上海）信息科技有限公司 手势识别方法、装置、设备和存储介质
CN109151202A (zh) 2018-08-30 2019-01-04 苏州矽典微智能科技有限公司 一种移动终端显示屏的控制方法及移动终端
CN109188414A (zh) 2018-09-12 2019-01-11 北京工业大学 一种基于毫米波雷达的手势运动探测方法
CN111134685B (zh) * 2018-11-02 2022-08-09 富士通株式会社 跌倒检测方法和装置
CN109583436B (zh) 2019-01-29 2023-09-12 杭州朗阳科技有限公司 一种基于毫米波雷达的手势识别系统
CN110035304B (zh) * 2019-03-08 2021-06-01 佛山市云米电器科技有限公司 一种应用于多空间的视频进度智能跟随播放方法及系统
US11442550B2 (en) * 2019-05-06 2022-09-13 Samsung Electronics Co., Ltd. Methods for gesture recognition and control
US10832418B1 (en) * 2019-05-09 2020-11-10 Zoox, Inc. Object velocity from images
CN110045370B (zh) 2019-05-10 2023-11-14 成都宋元科技有限公司 基于毫米波雷达的人体感知方法及其系统
CN110118966B (zh) 2019-05-28 2020-10-13 长沙莫之比智能科技有限公司 基于毫米波雷达的人员检测与计数系统
CN112050829B (zh) * 2019-06-06 2023-04-07 华为技术有限公司 一种运动状态确定方法及装置
CN110290353A (zh) * 2019-06-28 2019-09-27 Oppo广东移动通信有限公司 监控方法及装置、电子设备以及存储介质
US11327167B2 (en) * 2019-09-13 2022-05-10 Infineon Technologies Ag Human target tracking system and method
CN111104960B (zh) * 2019-10-30 2022-06-14 武汉大学 一种基于毫米波雷达和机器视觉的手语识别方法
US11567580B2 (en) * 2020-01-29 2023-01-31 Samsung Electronics Co., Ltd. Adaptive thresholding and noise reduction for radar data
US11513603B2 (en) * 2020-01-30 2022-11-29 Toyota Motor Engineering & Manufacturing North America, Inc. System and method for interpreting gestures
EP3869228B1 (en) * 2020-02-21 2022-07-06 Tata Consultancy Services Limited Edge based sensor actuation and control in iot network for event monitoring
CN111580060B (zh) * 2020-04-21 2022-12-13 北京航空航天大学 目标姿态识别的方法、装置和电子设备

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110181509A1 (en) * 2010-01-26 2011-07-28 Nokia Corporation Gesture Control
CN102509088A (zh) * 2011-11-28 2012-06-20 Tcl集团股份有限公司 一种手部运动检测的方法、装置及人机交互系统
CN103699220A (zh) * 2013-12-09 2014-04-02 乐视致新电子科技（天津）有限公司 一种根据手势运动轨迹进行操作的方法及装置
CN105607745A (zh) * 2016-03-16 2016-05-25 京东方科技集团股份有限公司 显示控制电路、显示控制方法和显示装置
US20190049558A1 (en) * 2017-08-08 2019-02-14 KC Sense Limited Hand Gesture Recognition System and Method
CN107961035A (zh) * 2017-12-29 2018-04-27 深圳开立生物医疗科技股份有限公司 一种超声探头以及控制超声诊断仪的方法和装置
CN108200706A (zh) * 2018-01-25 2018-06-22 宁波隔空智能科技有限公司 一种基于微波雷达手势识别技术的照明灯具及其控制方法
CN108344995A (zh) * 2018-01-25 2018-07-31 宁波隔空智能科技有限公司 一种基于微波雷达技术的手势识别装置及手势识别方法
CN109164915A (zh) * 2018-08-17 2019-01-08 湖南时变通讯科技有限公司 一种手势识别方法、装置、系统和设备
CN111624572A (zh) * 2020-05-26 2020-09-04 京东方科技集团股份有限公司 一种人体手部与人体手势识别的方法及装置

Also Published As

Publication number Publication date
CN111624572A (zh) 2020-09-04
US20230037445A1 (en) 2023-02-09
US11797098B2 (en) 2023-10-24
CN111624572B (zh) 2023-07-18

Similar Documents

Publication Publication Date Title
WO2021238710A1 (zh) 人体手部与手势识别的方法及装置、显示设备
US11340332B2 (en) Method and apparatus for processing radar data
US9411037B2 (en) Calibration of Wi-Fi localization from video localization
US6678394B1 (en) Obstacle detection system
Andriyenko et al. An analytical formulation of global occlusion reasoning for multi-target tracking
WO2021072710A1 (zh) 移动物体的点云融合方法、系统及计算机存储介质
CN106295459A (zh) 基于机器视觉和级联分类器的车辆检测和预警方法
Chang et al. People tracking with UWB radar using a multiple-hypothesis tracking of clusters (MHTC) method
Muresan et al. Multi-object tracking of 3D cuboids using aggregated features
CN112285698B (zh) 一种基于雷达传感器的多目标跟踪装置及方法
Aziz et al. Radar-camera fusion for road target classification
EP4050460A1 (en) Radar-based object tracking using a neural network
CN111999744A (zh) 一种无人机多方位探测、多角度智能避障方法
Javaid et al. A machine learning based method for object detection and localization using a monocular rgb camera equipped drone
Catalano et al. Uav tracking with solid-state lidars: dynamic multi-frequency scan integration
Diaz-Ruiz et al. Vision-only 3d tracking for self-driving cars
JP2021004874A (ja) 物品検出方法及び装置
TWI772208B (zh) 基於毫米波雷達的人數偵測方法
JP2004309166A (ja) 目標追尾装置
Tamas et al. Lidar and vision based people detection and tracking
WO2021078283A1 (zh) 一种数据关联方法及装置
Spraul et al. Persistent multiple hypothesis tracking for wide area motion imagery
CN114119465A (zh) 一种点云数据处理方法及装置
Zhang et al. Deep learning based multi-modal sensing for tracking and state extraction of small quadcopters
Chen et al. Research on optimization of visual navigation algorithm for real-time obstacle avoidance of uav moving target tracking

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: 21813085

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: 21813085

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 27/06/2023)

122 Ep: pct application non-entry in european phase

Ref document number: 21813085

Country of ref document: EP

Kind code of ref document: A1