• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    The present invention provides a method and system for recognizing a human posture on an electric bed. Real-time image information of a human body on the electric bed is acquired, the 5 electric bed is positioned, an approximate range of the human body is preliminarily positioned according to the position of the electric bed, and a human bounding box is recognized by using an RMPE posture recognition frame, and posture recognition is performed on the human body based on the human bounding box, and the sliding of a camera can be fixed by using the bed body before image acquisition to obtain a good human bounding box, so that the 10 human posture on the electric bed is recognized and positioned with high accuracy.
    公开号:NL2025724A
    申请号:NL2025724
    申请日:2020-06-02
    公开日:2021-05-28
    发明作者:Yuan Shuanghu；Li Wei；Cao Li；Liu Ning；Dong Yinjun；Zhang Benwu；Hu Xudong；Hou Wenhong；Zhu Shouhui；Yu Jinming
    申请人:Shandong Cancer Hospital And Inst；Univ Shandong；
    IPC主号:
    专利说明:

    METHOD AND SYSTEM FOR RECOGNIZING HUMAN POSTURE ON
    ELECTRIC BED Field of the Invention The present invention belongs to the technical field of human posture recognition, and specifically relates to a method and system for recognizing a human posture on an electric bed. Background of the Invention The statement of this section merely provides background art information related to the present invention, and does not necessarily constitute the prior art. In recent years, with the development of medical and health technology and computer technology, in order to reduce the workload of medical staff or family members in the family environment, electric beds are emerging. During the movement of an electric bed, it is very important for progress notes and patient safety to monitor the human body on the bed in real time and recognize the human posture through an intelligent terminal. At present, human posture recognition has been greatly developed in the fields of motion tracking, safety production, smart supervision, studio entertainment, judicial analysis, scenic spots, etc. The human posture recognition includes visual and non-visual processing methods. The visual human posture recognition method refers to analysis and processing of image data acquired by a camera, and the non-visual human posture recognition refers to posture recognition through acceleration measurement or sensors. For the immediate needs of posture recognition of electric beds, the non-visual human posture recognition method that requires a lot of calculations is not applicable. In view of the current situation that patients on electric beds need to wear therapeutic instruments or medicine bags, the non-visual human posture recognition method based on wearing single or multiple sensors is not applicable, because the human posture recognition on electric beds has extremely high real-time requirement, and the detection speed also needs to meet the real-time requirement while the accuracy is met.
    In addition, the existing visual posture recognition method is greatly affected by the quality of a human bounding box, needs to position a high-quality bounding box, is limited by the fixed position of the camera, often makes errors in positioning, and has redundant detection results.
    Summary of the Invention In order to solve the above problems, the present invention proposes a method and system for recognizing a human posture on an electric bed.
    The present invention can obtain a good human bounding box, so that the human posture on the electric bed is recognized and positioned with high accuracy.
    According to some embodiments, the present invention adopts the following technical solutions: A method for recognizing a human posture on an electric bed, including the following steps: acquiring real-time image information of a human body on the electric bed, and positioning the electric bed, preliminarily positioning an approximate range of the human body according to the position of the electric bed, and recognizing a human bounding box by using an RMPE posture recognition frame; and performing posture recognition on the human body based on the human bounding box.
    As a further limitation, a rotatable camera is disposed on the electric bed in a sliding manner, and the image information is acquired by the camera.
    As a further limitation, the size and range of the field of view of the camera are adjusted by sliding, rotating and focusing, and the camera acquires a human image in real time during the recognition process.
    As a further limitation, the process of acquiring image information includes: framing the position of the electric bed in the screen by using the camera; and checking the shooting screen in real time and controlling the slidable camera to avoid any background except the electric bed.
    As a further limitation, the process of positioning the human bounding box includes: representing the approximate range of the human body based on the stored size parameters of the electric bed; performing recognition through the RMPE posture recognition frame; and taking the approximate range frame and the RMPE posture recognition frame to obtain an intersection of the two frames, thus obtaining the human bounding box.
    As a further limitation, the process of performing posture recognition on the human body includes: performing symmetric spatial transformer network (SSTN) transformation, receiving human candidate boxes by using an STN, and generating candidate postures by an SDTN; performing posture recognition on the human body in the bounding box by using a Stacked Hourglass algorithm network; performing, after the recognition is completed, inverse SSTN transformation to return to the original image; and eliminating posture redundancy by using a posture non-maximum suppression (NMS) method, and selecting the one with the highest confidence from the redundant candidate boxes as a reference to obtain a posture vector.
    A system for recognizing a human posture on an electric bed, including an electric bed and a processor, wherein: a rotatable camera is disposed on the electric bed in a sliding manner, and image information is acquired by the camera; the processor receives the acquired image information and positions the electric bed accordingly, preliminarily positions an approximate range of a human body according to the position of the electric bed, recognizes a human bounding box by using an RMPE posture recognition frame, and performs posture recognition on the human body based on the human bounding box.
    As a further limitation, a slide bar is disposed on the electric bed, a slider is connected to the slide bar in a sliding manner, a holder is connected to the slider, and the camera is disposed on the holder.
    A computer-readable storage medium, storing multiple instructions adapted to be loaded by a processor of a terminal device to execute the method for recognizing a human posture on an electric bed.
    A terminal device, including a processor and a computer-readable storage medium, wherein the processor is used to implement various instructions, and the computer-readable storage medium is used to store multiple instructions, and the instructions are adapted to be loaded by the processor to execute the method for recognizing a human posture on an electric bed.
    Compared with the prior art, the beneficial effects of the present invention are: By combining the characteristics that the body of the electric bed has a fixed and known size and the human posture recognition object is a separate patient on the bed, and using a slidable camera, the present invention meets the requirements of accuracy and timeliness, and avoids the disadvantage that the fixed field of view of the human posture recognition camera is limited.
    In consideration of the differences between different individuals on the bed, if a standard predefined human body boundary is not specified in advance, due to the differences in height and weight of the individuals, a large number of neural network trainings are required to increase the algorithms, whereas an interface directly locking the bed is used to reduce the workload and improve the speed.
    The method of RMPE is combined into the SPPE single-person detection, and at the same time, the sliding of the camera can fixed by using the bed body before image acquisition to obtain a good human bounding box, so that the human posture on the electric bed 1s recognized and positioned with high accuracy.
    In order to avoid redundant detection results, the present invention uses NMS to eliminate posture redundancy so as to obtain a posture vector with the highest confidence.
    The combination of the posture recognition frame and the Stacked Hourglass algorithm network can well implement posture recognition.
    Brief Description of the Drawings The accompanying drawings constituting a part of the present invention are used for providing a further understanding of the present invention, and the schematic 5 embodiments of the present invention and the descriptions thereof are used for interpreting the present invention, rather than constituting improper limitations to the present invention.
    Fig. 1 is a flowchart of a method for recognizing a human posture on an electric bed; Fig. 2 is a diagram showing the placement position of a camera and the movable adjustment of field-of-view components; Fig. 3 explains a flow of processing posture redundancy in human posture recognition on the electric bed; Fig. 4 1s a schematic diagram of a human posture recognition process on the electric bed; Fig. Sis a result diagram of human posture recognition on the electric bed.
    Detailed Description of the Embodiments The present invention will be further illustrated below in conjunction with the accompanying drawings and embodiments.
    It should be pointed out that the following detailed descriptions are all exemplary and aim to further illustrate the present invention. Unless otherwise specified, all technical and scientific terms used in the descriptions have the same meanings generally understood by those of ordinary skill in the art of the present invention.
    It should be noted that the terms used herein are merely for describing specific embodiments, but are not intended to limit exemplary embodiments according to the present invention. As used herein, unless otherwise explicitly pointed out by the context, the singular form is also intended to include the plural form. In addition, it should also be understood that when the terms “include” and/or “comprise” are used in the specification, they indicate features, steps, operations, devices, components and/or their combination.
    A method for recognizing a human posture on an electric bed includes the following steps: A, acquiring real-time image information of a human body on the electric bed by a slidable camera; B, positioning a human bounding box; and C, performing posture recognition by using a human image; The method of acquiring the image information by the slidable camera in step A includes: Al, framing the position of the electric bed in the screen by the slidable camera on the electric bed; and A2, checking the shooting screen in real time and controlling the slidable camera to avoid any background as much as possible except the electric bed.
    The method of positioning the human bounding box in step B includes: B1, storing, because the electric bed is special and the human body boundary does not exceed the electric bed, parameters of the electric bed in a memory, including factors such as length, width and height within a two-dimensional plane, that is, framing the bed to represent an approximate range of the human body; B2, performing recognition through an RMPE posture recognition frame; and B3, taking an approximate range frame and RMPE by using IoU to obtain an intersection of the frames, thus obtaining the high-quality human bounding box.
    The method of performing human posture recognition in step C includes: CL, performing SSTN transformation, consisting of STN and SDTN, wherein STN receives human candidate boxes, and SDTN generates candidate postures; C2, performing posture recognition on the human body in the bounding box by using a Stacked Hourglass algorithm network; C3, performing, after the recognition is completed, inverse SSTN transformation to return to the original image; and C4, eliminating posture redundancy by using NMS, and selecting the one with the highest confidence from the redundant candidate boxes as a reference to obtain a posture vector.
    Embodiment 1: As shown in Fig. 1, a method for recognizing a human posture on an electric bed includes the following steps: Step S100, adjusting the field of view of a camera; Step S200, acquiring a real-time image of a human body on the electric bed; Step S300, extracting bed information in a memory, generating a predefined human body frame, then generating an RMPE posture recognition frame, and performing IoU processing, as shown in Fig. 4, to obtain an intersection of the two frames, thus obtaining a human bounding box; and Step S400, performing posture recognition by using a Stacked Hourglass algorithm network and an RMPE method to obtain a final result shown in Fig. 5. Each step will be described and explained in detail below. In steps S100 and S200, the size and range of the field of view of the camera used in the present invention are adjusted by sliding, rotating and focusing, the camera can acquire a human image in real time during the recognition process, and the two steps can increase the size of subsequent IoU. The following description will be made with reference to Fig. 2. (D is a slide bar and a slider, which ensure that the camera can be slide left and right within the plane shown in the figure; (2) is a rotation joint of the camera, which can ensure that the camera is rotated up and down within the plane shown to adjust the field of view; The following description will be made with reference to Fig. 3. At the beginning, the unadjusted field of view is too large, the bed is only detected by half, a mural and a table lamp are visible, and the camera can capture human postures beyond the bed within this range, so the field of view of the camera needs to be adjusted by moving the slider and the rotation joint to detect all the bed till other human bodies are hardly seen in the image within the range. In step S300, algorithm processing is performed under the predefined human body frame of the memory.
    The following describes the process of extracting the human bounding box:
    First, a high-quality artificial proposal is extracted by using STN.
    Mathematically, STN performs two-degree-of-freedom transformation.
    The formula is defined as (1): x XL (2) = 16,96, +) 0) 1 Where 8,, 8, and 63 are vectors. x , yf, x}, and y} are coordinates before and after the transformation, respectively.
    After single-person posture recognition is complete, SDTN is required to remap the human posture back to the original image coordinates.
    SDTN calculates y for removing the transformation, and generates grids based on y.
    The formula is defined in equation (2): xt x; | ) = [y1727s] (+) (2) Yi 1 Where y,;, 2, and y3 are inverse transform vectors, and the formulas are defined as (3) and (4). [vive] = [6,6,]7" (3) vs = 1x [117.1% (4). Embodiment 2: A system for recognizing a human posture on an electric bed includes an electric bed and a processor, wherein: A slide bar is disposed on the electric bed, a slider is connected to the slide bar in a sliding manner, a holder is connected to the slider, a camera is disposed on the holder, and image information is acquired by the camera; The processor receives the acquired image information and positions the electric bed accordingly, preliminarily positions an approximate range of a human body according to the position of the electric bed, recognizes a human bounding box by using an RMPE posture recognition frame, and performs posture recognition on the human body based on the human bounding box.
    Embodiment 3: A computer-readable storage medium stores multiple instructions adapted to be loaded by a processor of a terminal device to execute the method for recognizing a human posture on an electric bed.
    Embodiment 4: A terminal device includes a processor and a computer-readable storage medium, wherein the processor is used to implement various instructions; and the computer-readable storage medium is used to store multiple instructions, and the instructions are adapted to be loaded by the processor to execute the method for recognizing a human posture on an electric bed.
    A person skilled in the art should understand that the embodiments of the present invention may be provided as a method, a system, or a computer program product.
    Therefore, the present invention may be in the form of a full hardware embodiment, a full software embodiment, or an embodiment combining software and hardware.
    In addition, the present invention may be in the form of a computer program product implemented on one or more computer available storage media (including but not limited to a disk memory, a CD-ROM, an optical memory, etc.) including computer available program codes.
    The present invention is described with reference to flowcharts and/or block diagrams of the method, device (system), and the computer program product in the embodiments of the present invention.
    It should be understood that computer program instructions can implement each process and/or block in the flowcharts and/or block diagrams and a combination of processes and/or blocks in the flowcharts and/or block diagrams.
    These computer program instructions may be provided to a general-purpose computer, a dedicated computer, an embedded processor, or a processor of other programmable data processing device to generate a machine, so that a device configured to implement functions specified in one or more processes in the flowcharts and/or one or more blocks in the block diagrams is generated by using instructions executed by the general-purpose computer or the processor of other programmable data processing device.
    These computer program instructions may also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a product including an instruction device, where the instruction device implements functions specified in one or more processes in the flowcharts and/or one or more blocks in the block diagrams.
    These computer program instructions may also be loaded into a computer or other programmable data processing device, so that a series of operation steps are performed on the computer or other programmable data processing device to generate processing implemented by a computer, and instructions executed on the computer or other programmable data processing device provide steps for implementing functions specified in one or more processes in the flowcharts and/or one or more blocks in the block diagrams.
    Described above are merely preferred embodiments of the present invention, and the present invention is not limited thereto.
    Various modifications and variations may be made to the present invention for those skilled in the art.
    Any modification, equivalent substitution or improvement made within the spirit and principle of the present invention shall fall into the protection scope of the present invention.
    Although the specific embodiments of the present invention are described above in combination with the accompanying drawing, the protection scope of the present invention is not limited thereto.
    It should be understood by those skilled in the art that various modifications or variations could be made by those skilled in the art based on the technical solution of the present invention without any creative effort, and these modifications or variations shall fall into the protection scope of the present invention.
    权利要求:
    Claims (10)
    [1]
    A method of recognizing a human posture on an electric bed, comprising the steps of: acquiring real-time image information of a human body on the electric bed, and positioning the electric bed; preliminary positioning of an approximate range of the human body according to the position of the electric bed, and recognition of a human bounding box by means of an RMPE posture recognition frame; and performing a posture recognition on the human body based on the human bounding box.
    [2]
    The method of recognizing a human posture on an electric bed according to claim 1, wherein a rotatable camera on the electric bed is arranged in a sliding manner, and the image information is collected by means of the camera.
    [3]
    The method of recognizing a human posture on an electric bed according to claim 1, wherein the size and range of the field of view of the camera is adjusted by shifting, rotating and focusing, and the camera captures a human image in real time during the recognition process amasses.
    [4]
    A method of recognizing a human posture on an electric bed according to claim 1, wherein the process of collecting image information comprises: framing the position of the electric bed in the screen by means of the camera; and real-time recording screen control and sliding camera control to avoid any background outside the electric bed.
    [5]
    The method of recognizing a human posture on an electric bed according to claim 1, wherein the process of positioning a human bounding box comprises: representing the approximate range of the human body based on the stored size parameters of the electric bed; performing a recognition by means of the RMPE posture recognition frame; and use the frame of approximate range and the
    RMPE attitude recognition frame in order to obtain an intersection of the two frames and thus the human bounding box.
    [6]
    The method of recognizing a human posture on an electric bed according to claim 1, wherein the process of performing a posture recognition on the human body comprises: performing an SSTN (symmetric spatial transformer network) transformation, receiving human candidates bounding boxes by means of an STN, and generating candidates for attitudes by means of an SDTN; performing a posture recognition on the human body in the bounding box by means of a network of the type Stacked Hourglass algorithm network; performing, after the recognition is finished, a reverse SSTN transformation to return to the original image; and eliminating attitude redundancies by NMS, and selecting the candidate with the highest confidence among the redundant bounding box candidates as reference to obtain an attitude vector.
    [7]
    A system for recognizing a human posture on an electric bed, comprising an electric bed and a processor, wherein: a rotatable camera on the electric bed is arranged in a sliding manner, and image information is collected by the processor; the processor receives the acquired image information and positions the electric bed accordingly, prepositions an approximate range of a human body according to the position of the electric bed, and recognizes a human bounding box by means of an RMPE posture recognition frame, and a posture recognition on the human body based on the human bounding box.
    [8]
    The human posture recognition system on an electric bed according to claim 7, wherein a slide bar is arranged on the electric bed, a slide member is connected to the slide bar in a sliding manner, a holder is connected to the slide member, and the camera is on the container is set up.
    [9]
    A computer-readable storage medium that stores a plurality of instructions and is adapted to be loaded with a processor of a terminal in order to perform the method of recognizing a human posture on an electric bed according to any one of claims 1 to 6.
    [10]
    A terminal device comprising a processor and a computer readable storage medium, the processor being used to execute various instructions; and the computer readable storage medium is used to store a plurality of instructions, and the instructions are adapted to be loaded by the processor in order to implement the method of recognizing a human posture on an electric bed according to any one of claims | up to 6.
    类似技术:
    公开号 | 公开日 | 专利标题
    US20210177124A1|2021-06-17|Information processing apparatus, information processing method, and computer-readable storage medium
    CN106022213A|2016-10-12|Human body motion recognition method based on three-dimensional bone information
    CN108646915B|2020-12-15|Method and system for controlling mechanical arm to grab object by combining three-dimensional sight tracking and brain-computer interface
    Richa et al.2012|Vision-based proximity detection in retinal surgery
    WO2015013022A1|2015-01-29|Method and computations for calculating an optical axis vector of an imaged eye
    Sun et al.2014|Real-time gaze estimation with online calibration
    Gerós et al.2016|Facegram–Objective quantitative analysis in facial reconstructive surgery
    NL2025724B1|2021-12-14|Method and system for recognizing human posture on electric bed
    KR100325365B1|2002-03-04|Gaze detection system
    JP2006285531A|2006-10-19|Detection device for eye direction, detecting method for eye direction, program for executing the same detecting method for eye direction by computer
    US9672621B2|2017-06-06|Methods and systems for hair transplantation using time constrained image processing
    CN106385533B|2019-04-26|Panoramic video control method and system
    Bosco et al.2017|Multiple coordinate systems and motor strategies for reaching movements when eye and hand are dissociated in depth and direction
    Nandikolla et al.2019|Hybrid bci controller for a semi-autonomous wheelchair
    WO2020013778A2|2020-01-16|Evaluation method for the hair transplant process using the image processing and robotic technologies and the system of the method
    Kyriakoulis et al.2008|Fuzzy vergence control for an active binocular vision system
    Schiphorst et al.2020|Video2report: a video database for automatic reporting of medical consultancy sessions
    Colombo et al.1995|Prototype of a vision-based gaze-driven man-machine interface
    Schneider et al.2008|Vision system for wearable and robotic uses
    Hoyoux et al.2016|A new computer vision-based system to help clinicians objectively assess visual pursuit with the moving mirror stimulus for the diagnosis of minimally conscious state
    Kapralov et al.2019|EEG-Based Brain-Computer Interface for Control of Assistive Devices
    Elsamnah et al.2014|Medical navigation based on coloured markers for image-guided surgery
    Ratsamee et al.2015|Object search framework based on gaze interaction
    HOLLANDER2021|ROBOTIC TELEOPERATION VIA MOTION AND GAZE TRACKING
    Han et al.2021|FPLP3D: Security robot for face recognition in the workplace environment using face pose detection assisted controlled FACE++ tool position: A three-dimensional robot
    同族专利:
    公开号 | 公开日
    NL2025724B1|2021-12-14|
    CN110638461A|2020-01-03|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    US20190205630A1|2017-12-29|2019-07-04|Cerner Innovation, Inc.|Methods and systems for identifying the crossing of a virtual barrier|
    KR101279503B1|2011-12-23|2013-06-28|김종진|Apparatus for correction of spine and body posture|
    CN106464844A|2014-04-08|2017-02-22|琼和欧文·雅各布斯以色列理工学院-康奈尔研究所|Systems and methods for configuring baby monitor cameras to provide uniform data sets for analysis|
    EP3143931B1|2014-05-13|2020-12-09|Omron Corporation|Posture estimation device and posture estimation method|
    JP6122188B1|2015-07-30|2017-04-26|ミネベアミツミ株式会社|Body condition detection device, body condition detection method, and bed system|
    CN205434140U|2015-12-25|2016-08-10|深圳市第二人民医院|Pancreas transplantation postoperative care sick bed|
    US20170316578A1|2016-04-29|2017-11-02|Ecole Polytechnique Federale De Lausanne |Method, System and Device for Direct Prediction of 3D Body Poses from Motion Compensated Sequence|
    CN108969242A|2018-05-24|2018-12-11|大连亿斯德环境科技有限公司|intelligent hospital bed|
    CN108764190B|2018-06-04|2021-09-24|山东财经大学|Video monitoring method for off-bed and on-bed states of old people|
    CN109124166A|2018-07-24|2019-01-04|深圳市赛亿科技开发有限公司|Intelligent mattress and its management-control method, electronic equipment and storage medium|
    法律状态:
    优先权:
    申请号 | 申请日 | 专利标题
    CN201910876708.6A|CN110638461A|2019-09-17|2019-09-17|Human body posture recognition method and system on electric hospital bed|
    [返回顶部]