Dynamic gesture recognition device
Through the wear mechanism and recognition mechanism of the dynamic gesture recognition device, the wear process of gesture recognition is simplified, the cumbersome wear problems in the prior art are solved, and the operation is simple and convenient and suitable for users of different arm sizes. When not in use, it is not allowed to use. Save space.
Patent Information
- Application Number
- CN202420647522.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Utility models(China)
- Current Assignee / Owner
- Filing Date
- 2024-03-29
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2034-03-29
AI Technical Summary
In the prior art, attitude sensors need to be provided at each joint of each finger, which leads to the user need to wear each attitude sensor at the corresponding position of the finger when using it, which is relatively cumbersome to wear.
Provided is a dynamic gesture recognition device, including a wearable mechanism and an identification mechanism, the wearable mechanism includes a fixed arm and a wearable part, and the identification mechanism includes a telescopic arm and a camera, and adjusts the camera by adjusting the angle and length between the telescopic arm and the fixed arm. To a position that matches the user's body shape, gesture recognition is realized.
It simplifies the wear process, is simple and convenient to operate, is suitable for users of different arm sizes, and is foldable when not in use to save space, improving the practicality and convenience of the device.
Smart Images

Figure CN222071253U_ABST
Abstract
Description
Technical Field
[0001] The utility model relates to a gesture action capture technology, in particular to a dynamic gesture recognition device. Background Art
[0002] With the continuous development of industrial technology and the popularization of automation technology, robots have become a necessary means to improve the efficiency of industrial production. In terms of robot control, teleoperated robots enable people to complete various tasks without being directly in dangerous environments or remote areas, and have broad application prospects. The movements of human hands contain a lot of motion information, and robots can be controlled through gesture recognition.
[0003] For example, patent CN 215298263 U discloses a gesture recognition device, which provides sixteen gesture sensors and collects data from each gesture sensor by a circuit board to form a gesture sensor matrix, thereby performing data compensation, so that the error range of the gesture recognition device becomes smaller.
[0004] However, since the patent requires that posture sensors be provided at each joint of each finger to detect the relative postures between the fingers and obtain corresponding posture parameters after analysis, the user needs to wear each posture sensor at the corresponding position of the finger when using it, which is relatively cumbersome to wear. Utility Model Content
[0005] The purpose of the utility model is to overcome the above-mentioned technical deficiencies and propose a dynamic gesture recognition device to solve the technical problem that in the prior art, a posture sensor needs to be set at each joint of each finger to detect the relative posture between each finger, resulting in the user needing to wear each posture sensor at the corresponding position of the finger when using it, which is relatively cumbersome to wear.
[0006] In order to achieve the above technical purpose, the utility model adopts the following technical solutions:
[0007] The utility model provides a dynamic gesture recognition device, comprising:
[0008] A wearing mechanism, comprising a fixed arm and a wearing portion, wherein the wearing portion is mounted on one end of the fixed arm and is used to be worn on the arm of a user; and
[0009] The identification mechanism includes a telescopic arm and a camera. The telescopic arm has a connecting end and a telescopic end located at both ends. The connecting end is rotatably connected to the fixed arm and can adjust the angle between the telescopic arm and the fixed arm during rotation. The distance between the telescopic end and the connecting end is adjustable. The camera is installed on the telescopic end.
[0010] In some embodiments, the connecting end is provided with a first rotating shaft, the first rotating shaft is provided with a first driven tooth, the end of the fixed arm away from the wearable portion is provided with a first shaft hole, and the first rotating shaft is rotatably disposed in the first shaft hole;
[0011] The wearing mechanism also includes a first motor and a first driving tooth. The first motor is installed on the fixed arm. The first driving tooth is arranged on the output shaft of the first motor and meshes with the first driven tooth.
[0012] In some embodiments, the camera is rotatably mounted on the telescopic end, and the axial direction of the camera is the same as the axial direction of the connecting end.
[0013] In some embodiments, the camera includes a base and a camera head, a mounting seat is provided on one side of the base, the mounting seat is provided with a second rotating shaft, the second rotating shaft is provided with a second driven tooth, the telescopic end is provided with a second shaft hole, the second rotating shaft is rotatably provided in the second shaft hole, and the camera head is installed on the base;
[0014] The identification mechanism further includes a second motor and a second driving tooth, wherein the second motor is mounted on the telescopic end, and the second driving tooth is mounted on the output shaft of the second motor and meshes with the second driven tooth.
[0015] In some embodiments, the camera includes a plurality of infrared cameras, and the plurality of infrared cameras are mounted on an end of the base facing away from the mounting base.
[0016] In some embodiments, the camera further comprises a plurality of infrared light emitters, and the plurality of infrared light emitters are mounted on an end of the base facing away from the base.
[0017] In some embodiments, the camera is a depth camera.
[0018] In some embodiments, the fixed arm and the telescopic arm are hollow cylindrical steel structures.
[0019] In some embodiments, the wearing portion is a wearing bandage connected to the end of the fixed arm, and Velcro is provided on the wearing bandage.
[0020] In some embodiments, the telescopic arm includes a connecting arm and a movable arm, the connecting arm is provided with a channel along its axial direction, one end of the movable arm is telescopically arranged in the channel, and the other end extends out of the channel to form the telescopic end, and the end of the connecting arm away from the movable arm is the connecting end;
[0021] The identification mechanism further comprises a hydraulic cylinder, which is mounted on the inner wall of the channel and has a hydraulic rod connected to the movable arm.
[0022] Compared with the prior art, when using the dynamic gesture recognition device provided by the utility model, the wearable part is first worn on the user's arm, and then the angle between the telescopic arm and the fixed arm is adjusted, and at the same time, the length of the telescopic arm is adjusted to adjust the camera at the telescopic end to a position that matches the user's body shape, so as to ensure that the camera can accurately and completely obtain the user's gestures, so that the user's gesture changes can be obtained through the camera, and then gesture recognition is facilitated to control the robot to work. In this way, it is only necessary to wear the wearable part on the user's arm and then adjust the position of the camera. The operation is simple and convenient, and it can be suitable for users with different arm sizes, and it is more practical. At the same time, when not in use, the telescopic arm can be rotated to a position folded with the fixed arm, saving space and facilitating storage and transportation. BRIEF DESCRIPTION OF THE DRAWINGS
[0023] Figure 1 It is a structural schematic diagram of an embodiment of the dynamic gesture recognition device provided by the utility model when it is worn on the arm of a user;
[0024] Figure 2 yes Figure 1 A structural diagram of a dynamic gesture recognition device;
[0025] Figure 3 for Figure 2 A structural schematic diagram of the dynamic gesture recognition device from another angle;
[0026] Figure 4 for Figure 1 Schematic diagram of the dynamic gesture recognition device;
[0027] Figure 5 for Figure 1 A control schematic diagram of a dynamic gesture recognition device;
[0028] Figure 6 for Figure 1 Flow chart of the dynamic gesture recognition device.
[0029] Description of reference numerals:
[0030] 100. Dynamic gesture recognition device; 1. Fixed arm; 2. Wearable part; 21. Wearable bandage; 3. Telescopic arm; 31. Connecting end; 32. Telescopic end; 33. Connecting arm; 34. Movable arm; 4. Camera; 41. Base; 42. Camera head; 43. Mounting base; 5. Arm. DETAILED DESCRIPTION
[0031] In order to make the purpose, technical solution and advantages of the utility model more clear, the utility model is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the utility model and are not used to limit the utility model.
[0032] In order to solve the technical problem that in the prior art, a posture sensor needs to be provided at each joint of each finger to detect the relative posture of each finger, which causes the user to wear each posture sensor at the corresponding position of the finger when using it, and the wearing is relatively cumbersome, the utility model provides a dynamic gesture recognition device, which is simple and convenient to operate, has better practicality, and occupies less space during transportation and storage, thereby improving convenience.
[0033] See also Figures 1 to 3 , Figures 1 to 3 The structure diagram of a dynamic gesture recognition device 100 in an embodiment of the utility model includes a wearing mechanism and a recognition mechanism; the wearing mechanism includes a fixed arm 1 and a wearing part 2, the wearing part 2 is installed at one end of the fixed arm 1, and is worn on the user's arm 5; the recognition mechanism includes a telescopic arm 3 and a camera 4, the telescopic arm 3 has a connecting end 31 and a telescopic end 32 located at both ends, the connecting end 31 is rotatably connected to the fixed arm 1, and can adjust the angle between the telescopic arm 3 and the fixed arm 1 when rotating, the distance between the telescopic end 32 and the connecting end 31 is adjustable, and the camera 4 is installed at the telescopic end 32. It should be understood that in this embodiment, the camera 4 is installed at the telescopic end 32 to obtain the user's gesture.
[0034] When using the dynamic gesture recognition device 100 provided by the utility model, first wear the wearable part 2 on the user's arm 5, then adjust the angle between the telescopic arm 3 and the fixed arm 1, and adjust the length of the telescopic arm 3 at the same time, so as to adjust the camera 4 at the telescopic end 32 to a position that matches the user's body shape, and ensure that the camera 4 can accurately and completely obtain the user's gestures, so that the user's gesture changes can be obtained through the camera 4, and then gesture recognition is facilitated to control the robot to work. In this way, it is only necessary to wear the wearable part 2 on the user's arm 5 and then adjust the position of the camera 4. The operation is simple and convenient, and it can be applied to users with different arm 5 sizes, and it is more practical. At the same time, when not in use, the telescopic arm 3 can be rotated to a position folded with the fixed arm 1, saving space and facilitating storage and transportation.
[0035] In one embodiment, the connecting end 31 is provided with a first rotating shaft, the first rotating shaft is provided with a first driven tooth, the fixed arm 1 is provided with a first axial hole at one end away from the wearing part 2, and the first rotating shaft is rotatably arranged in the first axial hole; the wearing mechanism also includes a first motor and a first driving tooth, the first motor is installed on the fixed arm 1, the first driving tooth is provided on the output shaft of the first motor, and meshes with the first driven tooth.
[0036] In this embodiment, the first motor drives the first driving tooth to rotate. With the cooperation of the first driving tooth and the first driven tooth, the first rotating shaft can be driven to rotate axially around the first axial hole, thereby realizing the angle adjustment between the telescopic arm 3 and the fixed arm 1, and the structure is stable and reliable.
[0037] In one embodiment, the camera 4 is rotatably mounted on the telescopic end 32, and its axis direction is the same as the axis direction of the connecting end 31. Specifically, the camera 4 includes a base 41 and a camera head 42, a mounting seat 43 is provided on one side of the base 41, the mounting seat 43 is provided with a second rotating shaft, the second rotating shaft is provided with a second driven tooth, the telescopic end 32 is provided with a second shaft hole, the second rotating shaft is rotatably arranged in the second shaft hole, and the camera head 42 is mounted on the base 41; the identification mechanism also includes a second motor and a second driving tooth, the second motor is mounted on the telescopic end 32, the second driving tooth is mounted on the output shaft of the second motor, and meshes with the second driven tooth.
[0038] In this embodiment, the second driving tooth is driven to rotate by the first motor, and then with the cooperation of the second driving tooth and the second driven tooth, the second rotating shaft can rotate in the second axial hole. In this way, the camera 42 can be rotated relative to the telescopic arm 3, that is, the angle between the camera 42 and the telescopic arm 3 can be flexibly adjusted, so that the posture change of the camera 42 is more flexible.
[0039] In one embodiment, the camera 42 includes a plurality of infrared cameras 42, which are mounted on one end of the base 41 facing away from the mounting base 43. The camera 42 also includes a plurality of infrared light emitters, which are mounted on one end of the base 41 facing away from the base 41. At the same time, the camera 42 is a depth camera 42.
[0040] In this embodiment, a depth camera 42 is used to obtain a depth gesture image, improve the gesture acquisition accuracy, and thus improve the robot control accuracy. In addition, an infrared camera 42 and an infrared light emitter are provided correspondingly, so that the device has a wider application scenario. Specifically, in this solution, the infrared light emitter is an infrared LED.
[0041] In one embodiment, the fixed arm 1 and the telescopic arm 3 are hollow cylindrical steel structures.
[0042] In this embodiment, in view of the requirements of the assembly task, a hollow cylindrical steel lightweight structure design is used, which has the advantages of light weight, high load ratio, low cost, and stable structure. For example, the equipment has strong applicability in continuous remote operation.
[0043] It should be noted that, in other embodiments, the wearing portion 2 may be in the form of two arc-shaped and hinged half-ring plates. In this embodiment, the wearing portion 2 is a wearing bandage 21 connected to the end of the fixed arm 1, and the wearing bandage 21 is provided with Velcro.
[0044] In this embodiment, the wearing method of the bandage 21 can be adjusted individually according to the operator's personal situation to ensure that the device fits the body comfortably and does not affect the operator's natural movements and activities.
[0045] In one embodiment, see Figure 3 The telescopic arm 3 includes a connecting arm 33 and a movable arm 34. The connecting arm 33 is provided with a channel along its axial direction. One end of the movable arm 34 is telescopically arranged in the channel, and the other end extends out of the channel to form a telescopic end 32, and the end of the connecting arm 33 away from the movable arm 34 is a connecting end 31; the identification mechanism also includes a hydraulic cylinder, which is installed on the inner wall of the channel, and its hydraulic rod is connected to the movable arm 34.
[0046] In this embodiment, the movable arm 34 is extended and retracted by the extension of the hydraulic cylinder to adjust the distance between the connection end 31 and the extension end 32, so that the camera 4 can completely and accurately obtain the user's gesture information. The structure is stable and reliable, and the hydraulic cylinder is arranged in the channel to improve the compactness of the device. In addition, an embedded ball can be arranged between the movable arm 34 and the inner wall of the channel to facilitate the extension and retraction of the movable arm 34 relative to the connection arm 33.
[0047] In order to better understand the present invention, the following Figures 4 to 6 The technical solution of the utility model is described in detail:
[0048] A right-handed spatial rectangular coordinate system is established with the direction of the human forearm along the hand as the positive direction of the X-axis. The telescopic arm 3 rotates in the XOZ plane with the connecting end 31 as the center. The base 41 of the camera 42 can rotate in the XOZ plane with the telescopic end 32 as the center. The telescopic arm 3 can be extended and retracted forward and backward along the X-axis direction.
[0049] The direction of the camera 42 during movement is in the XOZ plane, so that the camera 42 can recognize the human hand from bottom to top, thereby obtaining more human hand gesture information by recognizing the palm. The position and recognition angle of the telescopic arm 3 and the camera 42 can be adjusted to improve the accuracy and sensitivity of gesture recognition.
[0050] The basic principle of the utility model is that after the power supply is connected, the camera 42 (Leap Motion) captures the current human hand posture data, uses deep learning to analyze and identify gestures, obtains robot hand images, such as single-pointing, single-finger left rotation, single-finger right rotation, palm up, palm down and other gesture images, and sends the recognized gestures to the single-chip microcomputer for real-time preprocessing, adopts interference compensation, inputs the preprocessed tremor, jitter and other interference signals into their opposite values through the feedback link, and offsets the actual interference, thereby achieving high anti-interference ability. And match the recognized gesture image with the gesture in the gesture library, and then output it to the robot hand, so that it executes the corresponding action instructions.
[0051] The gesture recognition flow chart is as follows Figure 6 As shown in Figure 1. First, a depth camera is used to acquire image and depth information, and an RGB-D camera is used to acquire color and depth maps. This device can provide depth information about the distance of an object, making it possible to distinguish between the foreground (such as a human hand) and the background.
[0052] Next, in the preprocessing stage, noise removal (such as using bilateral filters), image enhancement (such as histogram equalization), and possible alignment techniques between depth images and color images are used, and grayscale images are obtained to improve image quality and reduce the computational burden of subsequent algorithms. Then, a convolutional neural network (CNN) is constructed using the preprocessed image, and the hand area is segmented from the complex background using an existing gesture dataset.
[0053] After acquiring the image of the hand, the next step is to obtain information such as the position and speed of the key points of the hand. The solution first applies a convolutional neural network (CNN) to the preprocessed image for in-depth analysis, aiming to accurately segment the hand area from the complex background. This step involves using deep learning technology, especially the multi-level feature extraction capability of CNN, to identify the shape and outline of the hand in the image. The CNN model learns the various forms and postures of the hand by training the existing gesture dataset, thereby achieving high-precision hand area segmentation.
[0054] After obtaining a clear hand image, this solution uses advanced key point detection algorithms, such as HandTrack, to identify and track the key points of the hand, including but not limited to the fingertips and palms. The HandTrack algorithm uses a deep learning model to predict the 2D or 3D position of the key points of the hand and can track the movement trajectory of these key points in consecutive frames.
[0055] By comparing the positions of key points between consecutive frames, the algorithm calculates the speed and direction of movement of each key point, providing the necessary dynamic information for subsequent gesture recognition. Next, using the position and movement information of the key points, this scheme implements gesture recognition through a hidden Markov model (HMM). HMM is a statistical model that assumes that the state of the system is hidden but can be inferred from the observed sequence (in this case, the movement sequence of the key points).
[0056] By training the HMM model to recognize specific gestures or actions in a predefined gesture database, this solution can accurately recognize a series of complex dynamic gestures. Ultimately, the recognized gesture information is used to remotely control the robot for precise robot control. This process involves converting the gesture recognition results into commands that the robot can understand, thereby controlling the robot to complete specified operations, such as moving, grasping, or performing specific tasks. In this way, this solution realizes an efficient and intuitive human-computer interaction method, expanding the application scenarios and control flexibility of remote-controlled robots.
[0057] The specific implementation methods of the utility model described above do not constitute a limitation on the protection scope of the utility model. Any other corresponding changes and modifications made according to the technical concept of the utility model should be included in the protection scope of the claims of the utility model.
Claims
1. A dynamic gesture recognition device, characterized in that: include: The wearing mechanism comprises a fixed arm and a wearing part, wherein the wearing part is mounted on one end of the fixed arm and is used to be worn on the user's arm; as well as The identification mechanism includes a telescopic arm and a camera. The telescopic arm has a connecting end and a telescopic end located at both ends. The connecting end is rotatably connected to the fixed arm and can adjust the angle between the telescopic arm and the fixed arm during rotation. The distance between the telescopic end and the connecting end is adjustable. The camera is installed on the telescopic end.
2. The dynamic gesture recognition device according to claim 1, characterized in that: The connecting end is provided with a first rotating shaft, the first rotating shaft is provided with a first driven tooth, the end of the fixed arm away from the wearable part is provided with a first shaft hole, and the first rotating shaft is rotatably arranged in the first shaft hole; The wearing mechanism also includes a first motor and a first driving tooth. The first motor is installed on the fixed arm. The first driving tooth is arranged on the output shaft of the first motor and meshes with the first driven tooth.
3. The dynamic gesture recognition device according to claim 1 or 2, characterized in that: The camera is rotatably mounted on the telescopic end, and the axis direction of the camera is the same as the axis direction of the connecting end.
4. The dynamic gesture recognition device according to claim 3, characterized in that: The camera comprises a base and a camera head, a mounting seat is provided on one side of the base, the mounting seat is provided with a second rotating shaft, the second rotating shaft is provided with a second driven tooth, the telescopic end is provided with a second shaft hole, the second rotating shaft is rotatably arranged in the second shaft hole, and the camera head is installed on the base; The identification mechanism further includes a second motor and a second driving tooth. The second motor is mounted on the telescopic end. The second driving tooth is mounted on the output shaft of the second motor and meshes with the second driven tooth.
5. The dynamic gesture recognition device according to claim 4, characterized in that: The camera comprises a plurality of infrared cameras, and the plurality of infrared cameras are mounted on one end of the base facing away from the mounting base.
6. The dynamic gesture recognition device according to claim 5, characterized in that: The camera head also includes a plurality of infrared light emitters, which are installed at one end of the base facing away from the base.
7. The dynamic gesture recognition device according to claim 4, characterized in that: The camera is a depth camera.
8. The dynamic gesture recognition device according to claim 1, characterized in that: The fixed arm and the telescopic arm adopt hollow cylindrical steel structures.
9. The dynamic gesture recognition device according to claim 1, characterized in that: The wearing part is a wearing bandage connected to the end of the fixed arm, and Velcro is arranged on the wearing bandage.
10. The dynamic gesture recognition device according to claim 1, characterized in that: The telescopic arm comprises a connecting arm and a movable arm, wherein the connecting arm is provided with a channel along its axial direction, one end of the movable arm is telescopically arranged in the channel, and the other end extends out of the channel to form the telescopic end, and the end of the connecting arm away from the movable arm is the connecting end; The identification mechanism further comprises a hydraulic cylinder, which is mounted on the inner wall of the channel and has a hydraulic rod connected to the movable arm.
