WO2017080399A1 - 一种人脸位置跟踪方法、装置和电子设备 - Google Patents

一种人脸位置跟踪方法、装置和电子设备 Download PDF

Info

Publication number
WO2017080399A1
WO2017080399A1 PCT/CN2016/104491 CN2016104491W WO2017080399A1 WO 2017080399 A1 WO2017080399 A1 WO 2017080399A1 CN 2016104491 W CN2016104491 W CN 2016104491W WO 2017080399 A1 WO2017080399 A1 WO 2017080399A1
Authority
WO
WIPO (PCT)
Prior art keywords
face
area
frame picture
next frame
region
Prior art date
Legal status (The legal status 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 status listed.)
Ceased
Application number
PCT/CN2016/104491
Other languages
English (en)
French (fr)
Chinese (zh)
Inventor
王楠
杜志军
张宇
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Alibaba Group Holding Ltd
Original Assignee
Alibaba Group Holding Ltd
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
Priority to AU2016352215A priority Critical patent/AU2016352215B2/en
Application filed by Alibaba Group Holding Ltd filed Critical Alibaba Group Holding Ltd
Priority to KR1020187016518A priority patent/KR102150776B1/ko
Priority to EP16863576.1A priority patent/EP3376469A4/en
Priority to JP2018524732A priority patent/JP6873126B2/ja
Priority to SG11201803990YA priority patent/SG11201803990YA/en
Priority to MYPI2018000714A priority patent/MY196931A/en
Publication of WO2017080399A1 publication Critical patent/WO2017080399A1/zh
Priority to US15/977,576 priority patent/US10410046B2/en
Anticipated expiration legal-status Critical
Priority to PH12018501017A priority patent/PH12018501017A1/en
Priority to US16/561,918 priority patent/US10713472B2/en
Priority to US16/925,072 priority patent/US11003893B2/en
Priority to US17/316,487 priority patent/US11423695B2/en
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/223—Analysis of motion using block-matching
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/161—Detection; Localisation; Normalisation
    • G06V40/167—Detection; Localisation; Normalisation using comparisons between temporally consecutive images
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30196—Human being; Person
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30196—Human being; Person
    • G06T2207/30201—Face

Definitions

  • the present application belongs to the field of image information data processing, and in particular, to a face location tracking method, apparatus, and electronic device.
  • Face tracking can generally refer to the process of determining the trajectory and size of a face in a video or sequence of images. Face tracking has been of great significance and wide application in the fields of image analysis and recognition, image monitoring and retrieval, and instant video communication.
  • the process of the face tracking may mainly include finding a position of a face in the video.
  • the processing method of the face tracking method includes: performing face detection on each frame of the picture, that is, each frame can be regarded as a single picture, and then face detection is performed on each frame of the picture, thereby calculating Get the position of the face in each frame of the picture.
  • face tracking loss or detection is often caused by sudden changes in light or scene, interference from glare or metering, and rapid movement of faces. Wrong question. This often leads to discontinuity of the face tracking image during user video monitoring, video calling, etc., which does not achieve the effect of real-time smooth tracking, which greatly reduces the user experience, especially in terminal devices with lower processing performance. It is obvious.
  • the face tracking method in the prior art is also difficult to meet the needs of users who require high face tracking.
  • the face tracking method in the prior art especially in complex scenes such as strong light changes, light interference, and rapid face movement, still has the problem of face tracking loss or error, resulting in blurred face and face in the video picture.
  • the tracking screen is discontinuous, etc., which reduces the effect of face detection and tracking and the user experience.
  • the purpose of the present invention is to provide a method, a device, and an electronic device for tracking a face position, which can accurately locate a face region in a frame picture in a complex scene with strong light changes, light interference, and rapid face movement. Face tracking effect. At the same time, the problem of face tracking loss is solved, and the face tracking efficiency of the process is achieved, and the user experience is improved.
  • a face location tracking method, apparatus and electronic device provided by the present application are implemented as follows:
  • a face location tracking method comprising:
  • Detecting a face area of the next frame picture determining a face of the next frame picture according to a preset selection rule based on the detection result of the first pre-selected area and the face area of the next frame picture Location tracking results.
  • a face location tracking device comprising:
  • a detecting module configured to detect a face area of the frame picture
  • a prediction area calculation module configured to calculate, according to a face area of the current frame picture detected by the detection module, a prediction area in which a face appears in a next frame picture of the current frame picture;
  • a pre-selected area calculation module configured to search, in the prediction area, a first pre-selected area that has a similarity with the face area to meet a predetermined requirement
  • a tracking result selection module configured to determine, according to the detection result of the first frame image face region of the current frame image, the first frame image according to a preset selection rule, according to the first pre-selected region and the detection module Face location tracking results.
  • An electronic device for tracking a face position the electronic device being configured to include:
  • An information acquiring unit configured to acquire a frame picture to be processed
  • a processing unit configured to detect a face area of the frame picture; and configured to calculate, according to the detected face area of the current frame picture, a prediction area in which a face appears in a next frame picture of the current frame picture, and Searching, in the prediction area, a first pre-selected area that meets a predetermined requirement with the face area; and is further configured to: use, by the first pre-selected area and the detecting module, the next frame of the current frame picture
  • the detection result of the face area determines the face position tracking result of the next frame picture according to a preset selection rule
  • a display unit configured to display a face position tracking result obtained by the processing unit.
  • the face location tracking method, apparatus, and electronic device provided by the present application may predict a range of a prediction area in which a face appears in a next frame picture according to a face area in a current frame picture. Can then be within the range of the predicted area A pre-selected area of the face area where the similarity of the face area of the previous frame picture reaches a predetermined requirement (for example, the highest similarity) is found. In this way, the information of the pre-selected face obtained according to the previous frame picture can be obtained in the next frame picture of the current frame. Further, in the solution of the present application, the face area of the next frame picture may be detected.
  • the pre-selected area calculated according to the previous frame may be used as the The face area of the next frame. If the face area can be detected, at least two face areas of the next frame picture can then be selected according to a preset selection rule to select the required face area as the final face position tracking of the next frame picture. the result of. Therefore, in the present application, even if a complex scene such as a sudden change in light causes a current region to detect a face region, the obtained pre-selected region based on the face region prediction of the previous frame can be used to track the face position, thereby ensuring the person. Face detection, tracking continuity, improved face detection and tracking, and user experience using face tracking.
  • FIG. 1 is a flow chart of a method for tracking a face location tracking method according to an embodiment of the present application
  • FIG. 2 is a schematic diagram of determining a prediction area in which a face appears in a picture of a next frame according to a face area of a current frame picture;
  • FIG. 3 is a schematic diagram of the first pre-selected area found by the present application in the prediction area matching
  • FIG. 4 is a schematic diagram of a selection scenario for determining a face position tracking result provided by the present application.
  • FIG. 5 is a schematic diagram of further searching for the second pre-selected area in the embodiment of the present application.
  • FIG. 6 is a schematic structural diagram of a module of an embodiment of a face position tracking device according to the present application.
  • FIG. 7 is a schematic structural diagram of a module of an embodiment of the pre-selected area calculation module provided by the present application.
  • FIG. 8 is a schematic structural diagram of a module of another embodiment of the pre-selected area calculation module according to the present application.
  • FIG. 9 is a schematic structural diagram of a module of an embodiment of a tracking result selection module provided by the present application.
  • FIG. 10 is a schematic structural diagram of an embodiment of an electronic device for tracking a face position according to the present application.
  • FIG. 1 is a flow chart of a method for an embodiment of a face location tracking method according to the present application.
  • the present application provides method operational steps as illustrated in the following embodiments or figures, more or fewer operational steps may be included in the method based on routine or no inventive effort.
  • the order of execution of the steps is not limited to the execution order provided by the embodiment of the present application.
  • the actual implementation of the method or terminal product of the method may be performed sequentially or in parallel according to the method shown in the embodiment or the drawings (for example, an environment of parallel processor or multi-thread processing).
  • FIG. 1 An implementation of a face location tracking method provided by the present application is as shown in FIG. 1.
  • the method may include:
  • face position tracking is mostly used in video stream information processing by a camera device, such as a monitoring device, or a user using a mobile phone camera to capture video.
  • the face location tracking method provided by the present application may include, but is not limited to, information processing of a video stream.
  • the face drawing in the continuous picture or the film film digital information may still be Use the solution of the present application.
  • the application scenario in which the user uses the front camera of the mobile phone to perform self-photographing is described in the embodiment.
  • the face area in the current frame picture of the current video may be acquired first. Specifically, face detection of the video frame picture can be performed at the beginning of the video stream.
  • face detection is required for each frame of the image until a human face is detected.
  • information related to the face image such as an image color parameter, a size, a frame distance, and the like, can be obtained.
  • the position of the face may also be represented by a certain area range.
  • the commonly used representation may include using a rectangular frame to indicate the location area where the detected face is located.
  • the present application may provide an implementation manner of detecting a face region.
  • the obtaining a face region of the current frame image may include:
  • the face area of the current frame picture is detected and acquired by the Adaboost method of reducing the classification level.
  • the Adaboost is an iterative algorithm, and the main processing procedure may include training different classifiers (weak classifiers) for the same training set, and then combining the weak classifiers to form a stronger final classifier ( Strong Classifier).
  • the Adaboost algorithm can generally determine the weight of each sample based on whether the classification of each sample in each training set is correct and the accuracy of the last overall classification. Then, the modified new data set can be sent to the lower classifier for training, and finally the classifier obtained by each training is finally merged as the final decision classifier.
  • the classifier of the Adaboost cascade structure is usually composed of a series of series of classifiers.
  • the classifiers of the previous stages are relatively simple in structure, and the number of features used is also small, but the detection rate is high, and at the same time, the negative samples with large differences from the target can be filtered out as much as possible.
  • the later classifiers use more features and more complex structures to distinguish those negative samples that are similar to the target from the target object.
  • the number of classification stages may be appropriately reduced according to requirements, thereby reducing the calculation amount of face detection and achieving fast implementation. Face Detection.
  • the main subject is usually the photographer.
  • the face of the face is larger than the screen or display.
  • the face of the camera is often the main focus. Therefore, in order to more accurately track the face position and more in line with the expectation of the user's face position tracking, in another embodiment of the present application, only the face closest to the camera can be tracked during the face position tracking process. In the process, the largest face frame in the current frame picture can be taken as the face tracking object. Therefore, in another embodiment of the present application, the acquiring a face area of the current frame picture may include:
  • the area corresponding to the face with the largest picture area in the current frame is selected as the face area of the current frame picture.
  • a face area of a current frame picture of a video may be acquired.
  • S2 Determine, according to the face region of the current frame picture, a prediction area in which a face appears in a picture of the next frame of the current frame picture.
  • the prediction area where the face appears in the next frame picture may be determined according to the face area of the current frame picture.
  • the prediction area may include a range of regions in which a face appears in a next frame picture of the current frame determined according to a certain algorithm or rule calculation.
  • 2 is a schematic diagram of determining a prediction area in which a face appears in a next frame picture according to a face area of a current frame picture. As shown in FIG. 2, in the current frame picture N, the rectangular frame A is a face area for detecting the acquired current frame picture.
  • next frame picture (N+1) The rectangular frame B formed by extending the K pixels of the rectangular frame area length and width of the face area A in the picture N in the previous frame (ie, the current frame N described above) may be used as the next frame picture (N+ 1)
  • the predicted area in which the face appears may be determined in other manners, for example, by forming a rectangular frame formed by expanding the length and width of the rectangular frame A of the face area of the current frame picture by 1.5 times. Forecast area and other methods.
  • S3 Search, in the prediction area, a first pre-selected area that has a similarity with the face area to meet a predetermined requirement.
  • the search may be found in the prediction area to be similar to the face area acquired in the previous frame (ie, the current frame N in S2).
  • the first preselected area of the high face area is the face area acquired in the previous frame (ie, the current frame N in S2).
  • a template matching method may be provided to calculate the similarity of the face region in the first preselected area in the next frame picture.
  • the current frame picture face area may be used as the original template, and the boundary range of the face area may be set as the size of the moving window, and each window movement may form a new matching module.
  • the data of the new moving window area can be acquired each time one step is moved, and the similarity between the moving window area and the face area is calculated separately.
  • the specific method for calculating or judging the similarity between two regions may not be limited in the present application, and other methods for realizing the same or similar functions may still be applied in the implementation process of the present application.
  • the determining, according to the face area of the current frame picture, the prediction area in which the face appears in the next frame picture of the current frame picture may include:
  • S301 Traverse the prediction area according to the first moving step to obtain the comparison area of the face area.
  • the first step length can be set according to the processing speed or processing precision requirement of the actual face position tracking.
  • the value of the first moving step in the embodiment may be greater than or equal to two. pixel.
  • the calculation of similarity of different image regions can select corresponding calculation methods and calculation parameters according to different application scenarios or data processing requirements.
  • the similarity of the face region to the comparison region may be calculated based on colors, textures, gradients, and the like of images of different regions.
  • the present application provides an implementation manner for calculating the similarity.
  • the similarity dis of the face region and the comparison region may be calculated by using the following formula:
  • effctiveNum [min(width,maxX)-max(1,minX)]*[min(height,maxX)-max(1,minY)]
  • left ori , left des , top ori , top des can be respectively represented as the left boundary position of the face region, the left boundary position of the current comparison region, the upper boundary position of the face region, and the current ratio.
  • max(a, b) may not be represented as an item having a larger value of a and b
  • min(a, b) may be expressed as an item having a larger value of a
  • a predetermined requirement may be set, which may be used to filter out the comparison area in the prediction area that meets the prediction requirement.
  • the predetermined requirement may be set such that the similarity of the comparison area to the face area reaches 90% or more. Or, after the similarity is sorted, specify the comparison area within the percentage, for example, the first three comparison areas with the highest similarity.
  • the comparison area where the similarity reaches a predetermined requirement may be set to include:
  • FIG. 3 is a schematic diagram of the first pre-selected area found by the present application in the prediction area matching.
  • a face in the current frame picture N can be found in the prediction area B of the next frame picture (N+1).
  • the area A matches the first preselected area C that meets the requirements.
  • only the area most similar to the face area of the current frame picture in the next frame comparison area may be selected as the first pre-selection area. In this way, compared with selecting a plurality of pre-regions, the amount of data processing can be reduced, and the pre-regions can be filtered out more quickly, and the face processing speed can be improved.
  • the template matching method used in this embodiment can perform the addition and subtraction of the pixel gray value only in a certain area when calculating the first pre-selected area, and does not need to do excessive processing compared with other existing tracking algorithms. Storage, time and space complexity is low. The scope of application is wider, especially for low-end type mobile phones and monitoring devices with weak information and data processing capabilities, which can effectively reduce the amount of calculation and improve the face tracking accuracy. On the other hand, in the face tracking environment of close-range video shooting, for example, in the application scene of the front camera of the mobile phone, when the user is taking a self-portrait, the proportion of the screen occupied by the face is likely to be large.
  • the template matching method described in this embodiment can obtain more information about the effective area of the face region in the entire video interface, and the tracking result is more reliable than other tracking algorithms.
  • the first pre-selected area that meets the predetermined requirement that the similarity of the face area reaches the predetermined requirement may be searched according to a certain calculation method in the prediction area.
  • S4 detecting a face area of the next frame picture; determining, according to a detection result of the first pre-selected area and the face area of the next frame picture, the next frame picture according to a preset selection rule Face location tracking results.
  • next frame picture When switching to the next frame picture, it may be detected whether the next frame picture has a face area. If it is detected that the next frame picture has a face area, the next frame picture may obtain the detected face area and the face tracking prediction to obtain at least two face areas of the first preselected area. In the present application, the final face tracking result of the next frame picture may be obtained through a certain collaborative calculation analysis in combination with the at least two face regions.
  • the application may determine, according to a preset selection rule, which face area is selected as the final face position tracking result.
  • the selection rule described in this embodiment may include selecting the next one according to a different percentage of the area of the face area of the next frame picture and the first pre-selected area.
  • the frame picture face area is also a selection rule of the first pre-selected area.
  • the overlapping area of the face area of the next frame picture and the first pre-selected area may be defined as a percentage of the face area of the next frame picture or the area of the first pre-selected area. To coincide the coefficient Q.
  • FIG. 4 is a schematic diagram of a selection scenario for determining a face position tracking result provided by the present application.
  • the rectangular frame D may be represented as a detected face region of the next frame picture, which is referred to herein as a detection result; the rectangular frame C may represent tracking through the above steps S1 to S3 and other embodiments.
  • the calculated first pre-selected area of the next frame picture is referred to herein as a tracking result; the shaded part is the finalized face position tracking result of the next frame picture. If the detection result and the tracking result exist simultaneously in the next frame picture, if The detection result does not coincide with the tracking result, that is, the coincidence coefficient Q is 0, and the tracking result can be used as the face position tracking result, as shown by 4-1 in FIG.
  • the tracking result may be used as the face position tracking result, as shown in FIG. 4, 4-2. Shown. In another case, if the detection result overlaps with the tracking result in a large area and reaches a set coincidence requirement, for example, 95% of the areas overlap, the detection result may be selected as the face position tracking result, as shown in the figure. 4-4 shows. Certainly, if the face region is not detected in the next frame picture, the tracking result may be directly used as the face position tracking result, as shown by 4-4 in FIG. 4 .
  • the determining, according to the detection result of the face area of the next frame picture and the preset selection rule, the face position tracking result of the next frame picture may include:
  • the first pre-selected area is used as the face position tracking result of the next frame picture
  • the first preselected region is used as the face position tracking result of the next frame picture
  • the first preselected region is used as a face position tracking result of the next frame picture
  • the detected coincidence coefficient of the face area of the next frame picture and the first preselected area is greater than or equal to a predetermined threshold, the detected face area of the next frame picture is used as the The face position tracking result of the next frame picture.
  • This embodiment provides an implementation method of how to select the final face position tracking result from the detection result and the tracking result.
  • the embodiment of the present application can accurately and quickly track the face position in a complex environment such as rapid face movement, sudden light, and strong light interference.
  • the tracking of the face position can still be confirmed, and the effect of continuous tracking of the face position is achieved, so that the face tracking picture is smooth. Even if no frame loss occurs, a more suitable region can be selected between the detection result and the tracking result according to a predetermined selection rule as a face position tracking result in the frame picture, thereby improving the face tracking effect and improving the user experience.
  • the descriptions of the current frame, the next frame, the previous frame, and the previous array described in the present application can be considered as a relative concept for the description of the frame picture information processing object in an actual application.
  • the frame picture at a certain moment in the video stream can be marked as the current frame picture N
  • the corresponding next frame may be the N+1th frame picture
  • the previous frame may be the N-1th frame picture.
  • the process of tracking the face position in the N+2 frame picture may continue to be processed.
  • the current frame picture should be the N+1th frame picture.
  • the next frame picture of the current frame picture (N+1) may be the N+2 frame picture.
  • the processing result of the current frame picture may be used as the reference or initialization information of the next frame face position tracking to continue tracking the next frame picture.
  • the location of the face In some application scenarios, in general, the video stream needs to process more frame images per second, and can reach dozens or even dozens of frames.
  • the face position tracking processing if the face is lost due to strong light changes, the face moves quickly, etc., the face tracking loss is lost in a frame N, and the previous frame (N-1) can be detected or processed. The resulting face region is used as the face region of the face tracking lost frame N.
  • the detection of the previous frame (N-2) may be continued or the face region result may be obtained according to the processing. ,And so on.
  • the face is not detected in consecutive frames according to the preset judgment rule, it can be determined that the current face is no longer in the imaging range.
  • the value of the first moving step when searching for the prediction area in which the face appears in the next frame picture, the value of the first moving step may be set according to requirements, for example, it may be set to move 2 pixels each time or move 5 times each time. pixel. Generally, the larger the value of the moving step is set, the faster the speed of finding a similar area to the face of the previous frame, and the less the amount of data processed. The smaller the value of the moving step is set, the higher the accuracy of the corresponding search. In an embodiment in which the value of the first moving step is greater than or equal to two pixels, in order to further improve the accuracy of searching the first preselected region, another preferred embodiment provided by the present application is provided. The method may further include:
  • S304 In a second step range around the first pre-selected area, to find a second pre-selected area with the highest similarity to the face area, where the value of the second step is smaller than the first step.
  • the second preselected area obtained by the finer lookup can be used as the tracking result of the face position area of the next frame picture.
  • the detection result and the second preselected area may be selected, and therefore,
  • determining, according to the detection result of the face area of the first pre-selected area and the next frame picture, the face position tracking result of the next frame picture according to a preset selection rule comprises: The detection result of the second pre-selected area and the face area of the next frame picture is determined according to a preset selection rule, and the face position tracking result of the next frame picture is determined.
  • the first step is 2 pixels long, and in this embodiment, the person in the range of one pixel around the first preselected area and the person in the previous frame may be The face region is similarly calculated to obtain the region with the highest similarity value.
  • the similarity calculation method described in step S302 may be used to calculate the second pre-selected area, and of course, other methods for judging the similarity between the two areas are not excluded. The calculation method is not described here.
  • FIG. 5 is a schematic diagram of further searching for the second pre-selected area in the embodiment of the present application. As shown in FIG.
  • the rectangular frame C is a first pre-selected area of the face area determined by 2 pixels in the prediction area
  • the rectangular frame D is a ratio of one pixel point in the upper right corner of the first pre-selected area C.
  • the range of one pixel point around the first preselected area may include a comparison area C_d formed by moving the first preselected area downward by one pixel, a comparison area C_u formed by moving one pixel upward, and a lower left corner Move the alignment area C_ld formed by one pixel. Then, the similarity between the comparison area of one pixel point around the first preselected area and the face area may be separately calculated, and the comparison area with the highest similarity value may be selected as the second preselected area.
  • the first pre-selected area is obtained by calculating the first step of the value setting, which can effectively reduce the calculation amount of the picture comparison search and improve the data processing speed of the face position tracking. Then, in this embodiment, a finer lookup can be performed within a second step range in which the circumference of the first preselected area is smaller than the first step, and a second preselected area in which the tracking result is more accurate is obtained. In this way, the fast search processing can be achieved, and the accuracy of face tracking can be provided to improve the face tracking effect.
  • FIG. 6 is a schematic structural diagram of a module of a face position tracking device according to an embodiment of the present invention. As shown in FIG. 6, the device may include:
  • the detecting module 101 can be configured to detect a face area of the frame picture
  • the prediction area calculation module 102 may be configured to calculate, according to the face area of the current frame picture detected by the secondary detection module 101, a prediction area in which a face appears in a next frame picture of the current frame picture;
  • the pre-selected area calculation module 103 may be configured to search, in the prediction area, a first pre-selected area that has a similarity with the face area to meet a predetermined requirement;
  • the tracking result selection module 104 may be configured to determine the detection result according to the preset selection rule according to the detection result of the first pre-selected area and the detection module 101 for the next frame picture face area of the current frame picture. The face position tracking result of one frame of picture.
  • the detecting module 101 can continuously detect the face region in the captured frame image captured by the camera device along with the advancement of the time axis. For example, 15 video frames per second are captured in the video stream, and the face region in the current frame (Nth frame) can be detected when performing face position tracking. After detecting and tracking the information data of the current frame (Nth frame) picture, the face area in the next frame (N+1) picture can be continuously detected.
  • the Adaboost method for reducing the number of classification stages may be used to detect and acquire the face region of the current frame picture. In this way, the amount of data calculation of the face detection can be reduced, and the positioning and processing speed of the face position tracking can be improved.
  • the detecting module 101 detects a face area of a frame picture, including:
  • the area corresponding to the face with the largest picture area in the current frame is selected as the face area of the current frame picture.
  • FIG. 7 is a block diagram showing the structure of an embodiment of the pre-selected area calculation module 103 provided by the apparatus of the present application. As shown in FIG. 7, the pre-selected area calculation module 103 may include:
  • the comparison area module 1031 may be configured to traverse the prediction area according to the set first step length to obtain the comparison area of the face area;
  • the similarity calculation module 1032 can be configured to calculate the similarity between the face region and the comparison region.
  • the similarity calculation module 1032 may calculate the similarity dis of the face region and the comparison region by using the following formula:
  • effctiveNum [min(width,maxX)-max(1,minX)]*[min(height,maxX)-max(1,minY)]
  • left ori , left des , top ori , and top des are respectively represented as the left boundary position of the face region, the left boundary position of the current comparison region, the upper boundary position of the face region, and the current comparison.
  • the upper boundary position of the area width is the width of the face area, height is the height of the face area, and f(i, j) is represented as the coordinates of the current frame picture face area (i, j) the gray value of the pixel, g(i, j) is expressed as the gray value of the (i, j) pixel in the comparison region of the next frame;
  • x is the set empirical threshold, dis a similarity between the face region and the comparison region;
  • the first pre-selection module 1033 may be configured to use, as the first pre-selected area of the next frame picture, the comparison area in which the similarity in the next frame picture reaches a predetermined requirement.
  • the device described above and the calculation formula used in the similarity calculation module 1032 may be implemented in a computer-readable programming language, such as C language, in the specific implementation of the device/module. Or combined with the necessary hardware structure in hardware + software.
  • the preselection requirement set in the first preselection module 1033 may be set to be the highest similarity to the face region. Therefore, in another embodiment, the comparison area in the first pre-selection module 1033 that the similarity reaches a predetermined requirement may include:
  • the first step length set in the comparison area module 1031 may be set according to the requirements of the processing speed or accuracy of the face position tracking device according to the present application.
  • the value range of the first moving step may be set to be greater than or equal to two pixel points.
  • FIG. 8 is a block diagram showing another embodiment of a pre-selected area calculation module according to the present application. As shown in FIG. 8, the pre-selected area calculation module 103 may further include:
  • the second pre-selection module 1034 is configured to search, in a second step range around the first pre-selected area, a second pre-selected area with the highest similarity to the face area, where the second step is taken The value is less than the first step length.
  • the tracking result selection module 104 determines the detection result according to the preset selection rule according to the detection result of the first pre-selected area and the detection module 101 for the next frame picture face area of the current frame picture.
  • the result of the face position tracking of the next frame of the image includes: the detection result of the tracking result selection module 104 based on the second preselected area and the detection module 101 for the next frame picture face area of the current frame picture And determining a face position tracking result of the next frame picture according to a preset selection rule.
  • the result of the first pre-selected area may be searched in a finer manner within a second step size that is smaller than the first step, and the tracking result is more accurate. Second preselected area area. In this way, the fast search processing can be achieved, and the accuracy of the face position tracking can be provided to improve the face tracking effect.
  • FIG. 9 is a schematic structural diagram of a module of an embodiment of the tracking result selection module 104 provided by the present application. As shown in FIG. 9, the tracking result selection module 104 may include:
  • the detection calculation module 1041 may be configured to detect a face area of the next frame picture; and may be further configured to calculate the next frame face area and the said when detecting a face area of the next frame picture
  • the coincidence coefficient Q of the first preselected region may be represented by a face area of the next frame picture and a face area of the first pre-selected area occupying a face area of the next frame picture or the first The percentage of the area of a preselected area.
  • the selection module 1042 may be configured to: when the detection calculation module 1041 meets the face region where the next frame picture is not detected, the detection coefficient calculated by the detection calculation module 1041 is 0, and the detection calculation module 1041 calculates When the obtained coincidence coefficient is less than at least one of the predetermined thresholds, the first pre-selected area is used as the face position tracking result of the next frame picture; and may also be used for calculation by the detection calculation module 1041. When the coincidence coefficient is greater than or equal to the predetermined threshold, the face region of the next frame picture detected by the detecting module 101 is used as the face position tracking result of the next frame picture.
  • the solution of the embodiment provides a selection scheme for determining the final face position tracking result from the detection result and the tracking result.
  • the embodiment of the present application can accurately and quickly track the face position in a complex environment such as rapid face movement, sudden light, and strong light interference.
  • the tracking of the face position can still be confirmed, and the continuous tracking effect of the face position is achieved, so that the face tracking picture is smooth. Even if no frame loss occurs, a more suitable region can be selected between the detection result and the tracking result according to a predetermined selection rule as a face position tracking result in the frame picture, thereby improving the face tracking effect and improving the user experience.
  • the face location tracking method or device described in the present application can be applied to a variety of terminal devices to achieve faster, accurate, and process face location tracking, such as video capture of a mobile communication terminal based on Android or iOS system.
  • FIG. 10 is a schematic structural diagram of an embodiment of an electronic device for tracking a face position according to the present application. As shown in FIG. 10, the electronic device may be configured to include:
  • the information acquiring unit 1 can be configured to obtain a frame picture to be processed.
  • the processing unit 2 may be configured to detect a face area of the frame picture; and may be further configured to calculate, according to the detected face area of the current frame picture, a prediction area in which a face appears in a next frame picture of the current frame picture And searching, in the prediction area, a first pre-selected area that is similar to the face area to meet a predetermined requirement; and may be further configured to use the first pre-selected area and the detecting module to view the current frame picture a detection result of a face region of a frame of pictures, and determining a face position tracking result of the next frame picture according to a preset selection rule;
  • the display unit 3 can be used to display the face position tracking result obtained by the processing unit 2.
  • the information acquiring unit 1 in the electronic device described in the present application may include a front camera, a rear setting head of the mobile terminal, or an imaging device that monitors the setting.
  • a computer processes real-time or image-completed image information data may be included, for example, a process of tracking a piece of video information by a computer.
  • the processing unit 2 may include a central processing unit (CPU), and may of course include other single-chip microcomputers having logic processing capabilities, logic gate circuits, integrated circuits, and the like.
  • the display unit 3 may generally include a display, a display screen of a mobile terminal, a projection device, and the like.
  • the present application is not limited to the data that must be the standard or the data mentioned in the embodiment. Processing and information display.
  • the above description of the various embodiments in the present application is only an application in some embodiments of the present application.
  • the slightly modified processing method may also implement the foregoing embodiments of the present application. Program.
  • the same application can still be implemented in other non-innovative variations of the processing method steps described in the above embodiments of the present application, and details are not described herein again.
  • the unit or module illustrated in the above embodiments may be implemented by a computer chip or an entity, or by a product having a certain function.
  • the functions are divided into various modules and units respectively.
  • the functions of multiple modules may be implemented in the same software or software and/or hardware, such as the first pre-selected module and the second pre-selected module, in implementing the present application.
  • Modules that implement the same function can also be implemented by a combination of multiple submodules or subunits.
  • the controller can be logically programmed by means of logic gates, switches, ASICs, programmable logic controllers, and embedding.
  • the application can be described in the general context of computer-executable instructions executed by a computer, such as C language, or program modules based on Android, iOS design platforms, and the like.
  • program modules include routines, programs, objects, components, data structures, classes, and the like that perform particular tasks or implement particular abstract data types.
  • the present application can also be practiced in distributed computing environments where tasks are performed by remote processing devices that are connected through a communication network.
  • program modules can be located in both local and remote computer storage media including storage devices.
  • the present application can be implemented by means of software plus a necessary general hardware platform. Based on such understanding, the technical solution of the present application may be embodied in the form of a software product in essence or in the form of a software product, which may be stored in a storage medium such as a ROM/RAM or a disk. , an optical disk, etc., includes instructions for causing a computer device (which may be a personal computer, mobile terminal, server, or network device, etc.) to perform the methods described in various embodiments of the present application or portions of the embodiments.
  • a computer device which may be a personal computer, mobile terminal, server, or network device, etc.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Multimedia (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • General Health & Medical Sciences (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Human Computer Interaction (AREA)
  • Health & Medical Sciences (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Analysis (AREA)
  • Computational Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Algebra (AREA)
  • Operations Research (AREA)
  • Databases & Information Systems (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Image Analysis (AREA)
  • Studio Devices (AREA)
PCT/CN2016/104491 2015-11-12 2016-11-04 一种人脸位置跟踪方法、装置和电子设备 Ceased WO2017080399A1 (zh)

Priority Applications (11)

Application Number Priority Date Filing Date Title
MYPI2018000714A MY196931A (en) 2015-11-12 2016-11-04 Face location tracking method, apparatus, and electronic device
KR1020187016518A KR102150776B1 (ko) 2015-11-12 2016-11-04 얼굴 위치 추적 방법, 장치 및 전자 디바이스
EP16863576.1A EP3376469A4 (en) 2015-11-12 2016-11-04 Method and device for tracking location of human face, and electronic equipment
JP2018524732A JP6873126B2 (ja) 2015-11-12 2016-11-04 顔位置追跡方法、装置及び電子デバイス
SG11201803990YA SG11201803990YA (en) 2015-11-12 2016-11-04 Face location tracking method, apparatus, and electronic device
AU2016352215A AU2016352215B2 (en) 2015-11-12 2016-11-04 Method and device for tracking location of human face, and electronic equipment
US15/977,576 US10410046B2 (en) 2015-11-12 2018-05-11 Face location tracking method, apparatus, and electronic device
PH12018501017A PH12018501017A1 (en) 2015-11-12 2018-05-15 Method, apparatus, and electronic device for face tracking
US16/561,918 US10713472B2 (en) 2015-11-12 2019-09-05 Face location tracking method, apparatus, and electronic device
US16/925,072 US11003893B2 (en) 2015-11-12 2020-07-09 Face location tracking method, apparatus, and electronic device
US17/316,487 US11423695B2 (en) 2015-11-12 2021-05-10 Face location tracking method, apparatus, and electronic device

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201510772348.7 2015-11-12
CN201510772348.7A CN106709932B (zh) 2015-11-12 2015-11-12 一种人脸位置跟踪方法、装置和电子设备

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US15/977,576 Continuation US10410046B2 (en) 2015-11-12 2018-05-11 Face location tracking method, apparatus, and electronic device

Publications (1)

Publication Number Publication Date
WO2017080399A1 true WO2017080399A1 (zh) 2017-05-18

Family

ID=58694441

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/104491 Ceased WO2017080399A1 (zh) 2015-11-12 2016-11-04 一种人脸位置跟踪方法、装置和电子设备

Country Status (10)

Country Link
US (4) US10410046B2 (https=)
EP (1) EP3376469A4 (https=)
JP (1) JP6873126B2 (https=)
KR (1) KR102150776B1 (https=)
CN (1) CN106709932B (https=)
AU (1) AU2016352215B2 (https=)
MY (1) MY196931A (https=)
PH (1) PH12018501017A1 (https=)
SG (2) SG11201803990YA (https=)
WO (1) WO2017080399A1 (https=)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108334811A (zh) * 2017-12-26 2018-07-27 大唐软件技术股份有限公司 一种人脸图像处理方法及装置
CN109241345A (zh) * 2018-10-10 2019-01-18 百度在线网络技术（北京）有限公司 基于人脸识别的视频定位方法和装置
CN111429338A (zh) * 2020-03-18 2020-07-17 百度在线网络技术（北京）有限公司 用于处理视频的方法、装置、设备和计算机可读存储介质
JP2020534609A (ja) * 2018-03-06 2020-11-26 北京市商▲湯▼科技▲開▼▲発▼有限公司Ｂｅｉｊｉｎｇ Ｓｅｎｓｅｔｉｍｅ Ｔｅｃｈｎｏｌｏｇｙ Ｄｅｖｅｌｏｐｍｅｎｔ Ｃｏ．， Ｌｔｄ． 目標トラッキング方法及び装置、電子機器並びに記憶媒体
US20210217433A1 (en) * 2018-09-29 2021-07-15 Huawei Technologies Co., Ltd. Voice processing method and apparatus, and device
CN113853158A (zh) * 2019-07-22 2021-12-28 松下知识产权经营株式会社 步行功能评价装置、步行功能评价系统、步行功能评价方法、程序及认知功能评价装置
US11216955B2 (en) 2018-03-06 2022-01-04 Beijing Sensetime Technology Development Co., Ltd. Target tracking methods and apparatuses, electronic devices, and storage media

Families Citing this family (34)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106709932B (zh) 2015-11-12 2020-12-04 创新先进技术有限公司 一种人脸位置跟踪方法、装置和电子设备
CA3056026A1 (en) * 2016-03-22 2017-09-28 Communities Uncomplicated Inc. A method and system for tracking objects
CN106778585B (zh) * 2016-12-08 2019-04-16 腾讯科技（上海）有限公司 一种人脸关键点跟踪方法和装置
CN109508575A (zh) * 2017-09-14 2019-03-22 深圳超多维科技有限公司 人脸跟踪方法及装置、电子设备和计算机可读存储介质
CN107622252B (zh) * 2017-09-29 2022-02-22 百度在线网络技术（北京）有限公司 信息生成方法和装置
US10462370B2 (en) 2017-10-03 2019-10-29 Google Llc Video stabilization
CN107944381B (zh) * 2017-11-20 2020-06-16 深圳云天励飞技术有限公司 人脸跟踪方法、装置、终端及存储介质
CN108109107B (zh) * 2017-12-18 2021-08-20 北京奇虎科技有限公司 视频数据处理方法及装置、计算设备
CN108256479B (zh) * 2018-01-17 2023-08-01 百度在线网络技术（北京）有限公司 人脸跟踪方法和装置
US10171738B1 (en) * 2018-05-04 2019-01-01 Google Llc Stabilizing video to reduce camera and face movement
CN108776822B (zh) * 2018-06-22 2020-04-24 腾讯科技（深圳）有限公司 目标区域检测方法、装置、终端及存储介质
CN109064489A (zh) * 2018-07-17 2018-12-21 北京新唐思创教育科技有限公司 用于人脸追踪的方法、装置、设备和介质
CN110866428B (zh) * 2018-08-28 2023-12-15 杭州海康威视数字技术股份有限公司 目标跟踪方法、装置、电子设备及存储介质
CN110909579A (zh) * 2018-09-18 2020-03-24 杭州海康威视数字技术股份有限公司 一种视频图像处理方法、装置、电子设备及存储介质
KR102177453B1 (ko) * 2018-10-16 2020-11-11 서울시립대학교 산학협력단 얼굴 인식 방법 및 얼굴 인식 장치
CN109461169A (zh) * 2018-10-22 2019-03-12 同济大学 一种用于人脸追踪和人体定位的系统及方法
CN111127508B (zh) * 2018-10-31 2023-05-02 杭州海康威视数字技术股份有限公司 一种基于视频的目标跟踪方法及装置
KR102144101B1 (ko) * 2019-02-15 2020-08-12 주식회사 비플러스랩 안면 인식 방법 및 이를 적용한 출입문 안면 인식 시스템
TWI719409B (zh) * 2019-02-23 2021-02-21 和碩聯合科技股份有限公司 追蹤系統及其方法
CN109993081A (zh) * 2019-03-20 2019-07-09 浙江农林大学暨阳学院 一种基于道路视频及车牌检测的车流量统计方法
CN111722772A (zh) * 2019-03-21 2020-09-29 阿里巴巴集团控股有限公司 内容显示方法、装置及计算设备
CN110580444B (zh) * 2019-06-28 2023-09-08 时进制(上海)技术有限公司 一种人体检测方法及装置
SG10201913029SA (en) * 2019-12-23 2021-04-29 Sensetime Int Pte Ltd Target tracking method and apparatus, electronic device, and storage medium
CN111339855B (zh) * 2020-02-14 2023-05-23 睿魔智能科技（深圳）有限公司 基于视觉的目标跟踪方法、系统、设备及存储介质
CN111444875B (zh) * 2020-04-07 2024-05-03 珠海格力电器股份有限公司 一种人脸跟踪方法、装置、设备和计算机可读存储介质
CN111565300B (zh) * 2020-05-22 2020-12-22 深圳市百川安防科技有限公司 基于对象的视频文件处理方法、设备及系统
US11190689B1 (en) 2020-07-29 2021-11-30 Google Llc Multi-camera video stabilization
CN112417963A (zh) * 2020-10-20 2021-02-26 上海卫莎网络科技有限公司 一种优化视频目标检测、识别或分割的精度和效率的方法
CN114387279B (zh) * 2020-10-22 2026-03-17 北京三六零智领科技有限公司 人脸区域选取方法、设备、存储介质及装置
CN113936039B (zh) * 2021-10-15 2024-11-22 北京爱笔科技有限公司 一种对象跟踪方法、装置及系统
CN114220163B (zh) * 2021-11-18 2023-01-06 北京百度网讯科技有限公司 人体姿态估计方法、装置、电子设备及存储介质
CN116152872A (zh) * 2021-11-18 2023-05-23 北京眼神智能科技有限公司 人脸跟踪方法、装置、存储介质及设备
US12423835B2 (en) * 2021-12-17 2025-09-23 Samsung Electronics Co., Ltd. Method and apparatus with target tracking
US20240112410A1 (en) * 2022-09-29 2024-04-04 International Business Machines Corporation Learned feature prioritization to reduce image display noise

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060204036A1 (en) * 2005-03-09 2006-09-14 Dean Huang Method for intelligent video processing
CN101231703A (zh) * 2008-02-28 2008-07-30 上海交通大学 基于相关向量机和提升学习的多人脸跟踪方法
CN101527838A (zh) * 2008-03-04 2009-09-09 华为技术有限公司 对视频对象的反馈式对象检测与跟踪的方法和系统
CN103279751A (zh) * 2013-06-19 2013-09-04 电子科技大学 一种基于精确定位虹膜的眼动跟踪方法
CN103902960A (zh) * 2012-12-28 2014-07-02 北京计算机技术及应用研究所 一种实时人脸识别系统及其方法
CN104318211A (zh) * 2014-10-17 2015-01-28 中国传媒大学 一种抗遮挡人脸跟踪方法
CN104794439A (zh) * 2015-04-10 2015-07-22 上海交通大学 基于多相机的准正面人脸图像实时优选方法及系统

Family Cites Families (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20040228505A1 (en) * 2003-04-14 2004-11-18 Fuji Photo Film Co., Ltd. Image characteristic portion extraction method, computer readable medium, and data collection and processing device
EP1909229B1 (en) * 2006-10-03 2014-02-19 Nikon Corporation Tracking device and image-capturing apparatus
JP4987513B2 (ja) * 2007-03-07 2012-07-25 綜合警備保障株式会社 追跡装置、追跡方法、及び追跡プログラム
JP4882956B2 (ja) * 2007-10-22 2012-02-22 ソニー株式会社 画像処理装置および画像処理方法
US8131068B2 (en) * 2008-06-06 2012-03-06 Nikon Corporation Image matching device and camera
JP2010021943A (ja) * 2008-07-14 2010-01-28 Sanyo Electric Co Ltd 撮像装置
KR20100057362A (ko) * 2008-11-21 2010-05-31 삼성전자주식회사 영상의 유사도 판단 방법, 상기 방법을 기록한 기록 매체 및 상기 방법을 실행하는 장치
JP5127686B2 (ja) * 2008-12-11 2013-01-23 キヤノン株式会社 画像処理装置および画像処理方法、ならびに、撮像装置
JP5272789B2 (ja) * 2009-02-19 2013-08-28 株式会社ニコン 撮像装置
CN102054159B (zh) * 2009-10-28 2014-05-28 腾讯科技（深圳）有限公司 一种人脸跟踪的方法和装置
JP5355446B2 (ja) * 2010-02-19 2013-11-27 株式会社東芝 移動物体追跡システムおよび移動物体追跡方法
JP5746937B2 (ja) * 2011-09-01 2015-07-08 ルネサスエレクトロニクス株式会社 オブジェクト追跡装置
CN102306290B (zh) * 2011-10-14 2013-10-30 刘伟华 一种基于视频的人脸跟踪识别方法
US9626552B2 (en) * 2012-03-12 2017-04-18 Hewlett-Packard Development Company, L.P. Calculating facial image similarity
JP5955057B2 (ja) * 2012-03-30 2016-07-20 セコム株式会社 顔画像認証装置
US20140286527A1 (en) * 2013-03-20 2014-09-25 Qualcomm Incorporated Systems and methods for accelerated face detection
JP6117089B2 (ja) * 2013-12-12 2017-04-19 セコム株式会社 人物検出装置
CN106709932B (zh) 2015-11-12 2020-12-04 创新先进技术有限公司 一种人脸位置跟踪方法、装置和电子设备

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060204036A1 (en) * 2005-03-09 2006-09-14 Dean Huang Method for intelligent video processing
CN101231703A (zh) * 2008-02-28 2008-07-30 上海交通大学 基于相关向量机和提升学习的多人脸跟踪方法
CN101527838A (zh) * 2008-03-04 2009-09-09 华为技术有限公司 对视频对象的反馈式对象检测与跟踪的方法和系统
CN103902960A (zh) * 2012-12-28 2014-07-02 北京计算机技术及应用研究所 一种实时人脸识别系统及其方法
CN103279751A (zh) * 2013-06-19 2013-09-04 电子科技大学 一种基于精确定位虹膜的眼动跟踪方法
CN104318211A (zh) * 2014-10-17 2015-01-28 中国传媒大学 一种抗遮挡人脸跟踪方法
CN104794439A (zh) * 2015-04-10 2015-07-22 上海交通大学 基于多相机的准正面人脸图像实时优选方法及系统

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108334811A (zh) * 2017-12-26 2018-07-27 大唐软件技术股份有限公司 一种人脸图像处理方法及装置
JP2020534609A (ja) * 2018-03-06 2020-11-26 北京市商▲湯▼科技▲開▼▲発▼有限公司Ｂｅｉｊｉｎｇ Ｓｅｎｓｅｔｉｍｅ Ｔｅｃｈｎｏｌｏｇｙ Ｄｅｖｅｌｏｐｍｅｎｔ Ｃｏ．， Ｌｔｄ． 目標トラッキング方法及び装置、電子機器並びに記憶媒体
US11216955B2 (en) 2018-03-06 2022-01-04 Beijing Sensetime Technology Development Co., Ltd. Target tracking methods and apparatuses, electronic devices, and storage media
US11699240B2 (en) 2018-03-06 2023-07-11 Beijing Sensetime Technology Development Co., Ltd. Target tracking method and apparatus, and storage medium
US20210217433A1 (en) * 2018-09-29 2021-07-15 Huawei Technologies Co., Ltd. Voice processing method and apparatus, and device
CN109241345A (zh) * 2018-10-10 2019-01-18 百度在线网络技术（北京）有限公司 基于人脸识别的视频定位方法和装置
CN113853158A (zh) * 2019-07-22 2021-12-28 松下知识产权经营株式会社 步行功能评价装置、步行功能评价系统、步行功能评价方法、程序及认知功能评价装置
CN113853158B (zh) * 2019-07-22 2024-05-14 松下知识产权经营株式会社 步行功能评价装置、步行功能评价系统、步行功能评价方法、记录介质及认知功能评价装置
CN111429338A (zh) * 2020-03-18 2020-07-17 百度在线网络技术（北京）有限公司 用于处理视频的方法、装置、设备和计算机可读存储介质

Also Published As

Publication number Publication date
KR20180084085A (ko) 2018-07-24
JP2018533805A (ja) 2018-11-15
JP6873126B2 (ja) 2021-05-19
MY196931A (en) 2023-05-11
SG11201803990YA (en) 2018-06-28
US20200342211A1 (en) 2020-10-29
US10410046B2 (en) 2019-09-10
CN106709932A (zh) 2017-05-24
PH12018501017A1 (en) 2018-11-05
CN106709932B (zh) 2020-12-04
US11003893B2 (en) 2021-05-11
US20210264133A1 (en) 2021-08-26
US10713472B2 (en) 2020-07-14
US11423695B2 (en) 2022-08-23
AU2016352215B2 (en) 2020-10-01
EP3376469A4 (en) 2018-11-21
KR102150776B1 (ko) 2020-09-02
SG10202001825PA (en) 2020-04-29
US20190392199A1 (en) 2019-12-26
EP3376469A1 (en) 2018-09-19
AU2016352215A1 (en) 2018-06-21
US20180260614A1 (en) 2018-09-13

Similar Documents

Publication Publication Date Title
CN106709932B (zh) 一种人脸位置跟踪方法、装置和电子设备
US9179071B2 (en) Electronic device and image selection method thereof
CN109151375B (zh) 一种目标对象抓拍方法、装置及视频监控设备
CN108960163B (zh) 手势识别方法、装置、设备和存储介质
CN103093212B (zh) 基于人脸检测和跟踪截取人脸图像的方法和装置
US10007841B2 (en) Human face recognition method, apparatus and terminal
WO2019023921A1 (zh) 一种手势识别方法、装置及设备
US20160063731A1 (en) Tracking processing device and tracking processing system provided with same, and tracking processing method
WO2022206680A1 (zh) 图像处理方法、装置、计算机设备和存储介质
CN106165391A (zh) 增强的图像捕获
CN107784281B (zh) 人脸检测方法、装置、设备及计算机可读介质
EP2966591B1 (en) Method and apparatus for identifying salient events by analyzing salient video segments identified by sensor information
US11244154B2 (en) Target hand tracking method and apparatus, electronic device, and storage medium
WO2018228413A1 (zh) 一种目标对象抓拍方法、装置及视频监控设备
CN114463781A (zh) 确定触发手势的方法、装置及设备
WO2022206679A1 (zh) 图像处理方法、装置、计算机设备和存储介质
CN106030658B (zh) 用于确定视频的方位的方法及装置
JP6798609B2 (ja) 映像解析装置、映像解析方法およびプログラム
HK1237515B (zh) 一种人脸位置跟踪方法、装置和电子设备
HK1237515A (en) Face position tracking method and device and electronic equipment
HK1237515A1 (en) Face position tracking method and device and electronic equipment
Dai et al. Combining contrast saliency and region discontinuity for precise hand segmentation in projector-camera system
CN117635664A (zh) 无人机视频的单目标跟踪方法、装置、电子设备及存储介质

Legal Events

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

Ref document number: 16863576

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 11201803990Y

Country of ref document: SG

WWE Wipo information: entry into national phase

Ref document number: 2018524732

Country of ref document: JP

NENP Non-entry into the national phase

Ref country code: DE

WWE Wipo information: entry into national phase

Ref document number: 12018501017

Country of ref document: PH

ENP Entry into the national phase

Ref document number: 20187016518

Country of ref document: KR

Kind code of ref document: A

WWE Wipo information: entry into national phase

Ref document number: 1020187016518

Country of ref document: KR

WWE Wipo information: entry into national phase

Ref document number: 2016863576

Country of ref document: EP

ENP Entry into the national phase

Ref document number: 2016352215

Country of ref document: AU

Date of ref document: 20161104

Kind code of ref document: A