Gesture feature recognition method and apparatus

Abstract

The application discloses a gesture feature recognition method and device, and the method comprises the following steps: acquiring hand action information; performing calculation and processing on the hand action information to obtain hand posture information; the hand posture information comprises a plurality of target posture information; the target posture information represents the action feature condition of fingers and a palm; and target gesture feature information is determined based on the hand posture information. It can be seen that the application is beneficial to improving the gesture feature recognition efficiency and precision, and further improving the online real-time performance of gesture feature recognition and expanding the classification category of gesture action recognition.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent control, and in particular to a gesture feature recognition method and device. BACKGROUND

[0002] At present, the feature extraction method based on gesture original data statistical analysis is not suitable for different classifiers for different feature quantities, the recognition rate of the feature extraction method is relatively low, and the gesture action classification category cannot be too complex. Therefore, a gesture feature recognition method and device are provided to improve the gesture feature recognition efficiency and precision, and to improve the gesture feature recognition online real-time performance and expand the gesture action recognition classification category. SUMMARY

[0003] The technical problem to be solved by the present application is to provide a gesture feature recognition method and device to improve the gesture feature recognition efficiency and precision, and to improve the gesture feature recognition online real-time performance and expand the gesture action recognition classification category.

[0004] To solve the above technical problem, the first aspect of the embodiment of the present application discloses a gesture feature recognition method, which comprises:

[0005] Obtaining hand action information;

[0006] Calculating and processing the hand action information to obtain hand posture information; the hand posture information comprises a plurality of target posture information; the target posture information represents the action feature of fingers and palms;

[0007] Based on the hand posture information, target gesture feature information is determined.

[0008] As an optional implementation, in the first aspect of the embodiment of the present application, the calculating and processing of the hand action information to obtain hand posture information comprises:

[0009] The hand action information is normalized and calculated by using a normalization model to obtain normalized action data information; the normalized action data information comprises a plurality of normalized action information;

[0010] The normalization model is:

[0011] ;

[0012] In the formula, is a normalized action vector corresponding to the normalized action information; is a hand original data vector corresponding to the hand original data information in the hand action information; , , and are a first hand original value, a second hand original value, a third hand original value and a fourth hand original value distributed in the hand original data vector in turn respectively;

[0013] performing calculation processing on the normalized action data information to obtain hand posture information.

[0014] As an optional implementation, in the first aspect of the embodiment of the present application, the target posture information includes finger-palm posture information and wrist posture information; the normalized action information includes normalized finger action information, normalized palm action information and normalized wrist action information.

[0015] The performing calculation processing on the normalized action data information to obtain hand posture information includes:

[0016] For any normalized action information, performing calculation processing on the normalized finger action information and the normalized palm action information corresponding to the normalized action information to obtain the finger-palm posture information corresponding to the normalized action information.

[0017] Performing calculation processing on the normalized palm action information and the normalized wrist action information corresponding to the normalized action information to obtain the wrist posture information corresponding to the normalized action information.

[0018] As an optional implementation, in the first aspect of the embodiment of the present application, the performing calculation processing on the normalized finger action information and the normalized palm action information corresponding to the normalized action information to obtain the finger-palm posture information corresponding to the normalized action information includes:

[0019] Performing calculation processing on the normalized finger action information and the normalized palm action information corresponding to the normalized action information by using a first pose calculation model to obtain spatial pose information.

[0020] The first pose calculation model is:

[0021] ;

[0022] In the formula, the spatial pose information corresponding to the spatial pose vector is The normalized palm action information corresponds to a normalized palm action vector; The normalized finger action information corresponds to a normalized finger action vector;

[0023] Performing coordinate conversion processing on the spatial pose information by using a first coordinate conversion model to obtain the finger-palm posture information corresponding to the normalized action information.

[0024] ​The first coordinate conversion model is:

[0025] ;

[0026] In the formula, The finger and palm posture information corresponds to a finger and palm posture vector; The space pose vector.

[0027] As an optional implementation, in the first aspect of the embodiment of the application, the calculation and processing of the normalized hand palm action information and the normalized wrist action information corresponding to the normalized action information to obtain the wrist posture information corresponding to the normalized action information comprises:

[0028] Obtaining a rotation parameter;

[0029] The rotation parameter is calculated and processed to obtain normalized coordinate conversion quaternion information;

[0030] The normalized coordinate conversion quaternion information and the normalized hand palm action information corresponding to the normalized action information are multiplied to obtain palm coordinate conversion information;

[0031] The wrist posture information is obtained by calculating and processing the normalized wrist action information and the palm coordinate conversion information using a second pose calculation model;

[0032] The second pose calculation model is:

[0033] ;

[0034] In the formula, The wrist posture information corresponds to a wrist posture vector; The normalized wrist action information corresponds to a normalized wrist action vector; The palm coordinate conversion information corresponds to a palm coordinate conversion vector; The inverse matrix function is;

[0035] The wrist posture information is coordinate-converted to obtain the wrist posture information corresponding to the normalized action information.

[0036] As an optional implementation, in the first aspect of the embodiment of the application, the rotation parameter includes a first rotation angle and a second rotation angle; the first rotation angle and the second rotation angle are rotation angles relative to two directions perpendicular to each other;

[0037] The calculation and processing of the rotation parameter to obtain normalized coordinate conversion quaternion information comprises:

[0038] Carrying out sine and cosine calculation on the first rotation angle and the second rotation angle respectively to obtain first angle value information and second angle value information;

[0039] Filling the first angle value information and the second angle value information into preset first angle vectors and second angle vectors respectively to obtain first target angle vector information and second target angle vector information;

[0040] Carrying out cross multiplication calculation processing on the first target angle vector information and the second target angle vector information to obtain coordinate conversion vector information;

[0041] Carrying out normalization processing on the coordinate conversion vector to obtain normalized coordinate conversion quaternion information.

[0042] As an optional implementation, in the first aspect of the embodiment of the present application, the obtaining of the target gesture feature information based on the hand posture information comprises:

[0043] For any target posture information, data values in the target posture information are filled into a preset posture vector to obtain a target posture vector corresponding to the target posture information; the target posture vector comprises nine vector elements;

[0044] All the target posture vectors are sequentially arranged according to a data collection sequence of hand collection information in the hand action information to obtain a target posture matrix;

[0045] The target gesture feature information is determined based on the target posture matrix.

[0046] The second aspect of the embodiment of the present application discloses a gesture feature recognition device, which comprises:

[0047] An acquisition module is configured to acquire hand action information;

[0048] A processing module is configured to carry out calculation processing on the hand action information to obtain hand posture information; the hand posture information comprises a plurality of target posture information; the target posture information represents action feature conditions of fingers and palms;

[0049] A determination module is configured to determine target gesture feature information based on the hand posture information.

[0050] The third aspect of the present application discloses another gesture feature recognition device, which comprises:

[0051] A memory storing executable program codes;

[0052] A processor coupled with the memory;

[0053] The processor invokes the executable program code stored in the memory to execute part or all of the steps of the gesture feature recognition method disclosed in the first aspect of the embodiment of the present application.

[0054] The fourth aspect of the present application discloses a computer readable storage medium, which stores computer instructions, when the computer instructions are invoked, part or all of the steps of the gesture feature recognition method disclosed in the first aspect of the embodiment of the present application are executed.

[0055] Compared with the prior art, the embodiment of the present application has the following beneficial effects:

[0056] In the embodiment of the present application, hand action information is acquired, the hand action information is calculated and processed to obtain hand posture information, the hand posture information includes a plurality of target posture information, the target posture information represents the action feature of the fingers and the palm, and the target gesture feature information is determined based on the hand posture information. It can be seen that the present application is beneficial to improve the gesture feature recognition efficiency and precision, thereby improving the online real-time performance of gesture feature recognition and expanding the classification category of gesture action recognition. BRIEF DESCRIPTION OF DRAWINGS

[0057] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0058] Figure 1 is a flowchart of a gesture feature recognition method disclosed by the embodiment of the present application;

[0059] Figure 2 is a structural schematic diagram of a gesture feature recognition device disclosed by the embodiment of the present application;

[0060] Figure 3 is a structural schematic diagram of another gesture feature recognition device disclosed by the embodiment of the present application. DETAILED DESCRIPTION

[0061] In order to make the person skilled in the art better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings of the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0062] The terms "first", "second", and the like in the description and in the claims of the present application and above-described drawings are used to distinguish different objects, and are not used to describe a particular order. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, device, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but can optionally include other steps or units not listed, or can optionally include other steps or units inherent to the process, method, product, or apparatus.

[0063] Reference herein to "embodiments" means that the particular features, structures, or characteristics described in connection with the embodiments can be included in at least one embodiment of the present application. The appearance of the phrase in various places in the specification does not necessarily all refer to the same embodiment, nor is it necessarily independent or alternative to other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0064] The present application discloses a gesture feature recognition method and device, which is beneficial to improve the gesture feature recognition efficiency and accuracy, and further improve the online real-time performance of gesture feature recognition and expand the classification category of gesture action recognition. The following will be described in detail.

[0065] Embodiment one

[0066] Please refer to Figure 1 , Figure 1 is a flowchart of a gesture feature recognition method disclosed by the embodiments of the present application. Among them, Figure 1 The gesture feature recognition method described is applied to a warehouse management system, such as a local server or a cloud server for warehouse logistics gesture feature recognition management, which is not limited by the embodiments of the present application. As Figure 1 As shown in the figure, the gesture feature recognition method can include the following operations:

[0067] 101, obtain hand action information.

[0068] 102, calculate and process the hand action information to obtain hand posture information.

[0069] In the embodiments of the present application, the hand posture information includes a plurality of target posture information.

[0070] In the embodiments of the present application, the target posture information represents the action feature situation of the fingers and the palm.

[0071] 103, determine the target gesture feature information based on the hand posture information.

[0072] It should be noted that the gesture feature recognition method of the present application directly uses the hand freedom degree data as the feature quantity for gesture action recognition, which can well reflect the difference between different gesture actions, so as to improve the recognition rate, reduce the complexity of the classifier, and improve the online real-time operation capability.

[0073] It should be noted that the hand acquisition information in the hand action information is acquired based on the inertial sensor.

[0074] It can be seen that the gesture feature recognition method described in the embodiment of the present application is beneficial to improve the gesture feature recognition efficiency and accuracy, and further improve the online real-time performance of gesture feature recognition and expand the classification categories of gesture action recognition.

[0075] In an optional embodiment, the calculation and processing of the hand action information to obtain the hand posture information includes:

[0076] The hand action information is normalized and calculated by using the normalization model to obtain normalized action data information; the normalized action data information includes a plurality of normalized action information;

[0077] The normalization model is:

[0078] ;

[0079] In the formula, is the normalized action vector corresponding to the normalized action information; is the hand original data vector corresponding to the hand original data information in the hand action information; , , and are the first hand original value, the second hand original value, the third hand original value and the fourth hand original value distributed in the hand original data vector in sequence, respectively;

[0080] The normalized action data information is calculated and processed to obtain the hand posture information.

[0081] It should be noted that the hand original data vector corresponding element value is , , and .

[0082] It can be seen that the gesture feature recognition method described in the embodiment of the present application is beneficial to improve the gesture feature recognition efficiency and accuracy, and further improve the online real-time performance of gesture feature recognition and expand the classification categories of gesture action recognition.

[0083] In another optional embodiment, the target gesture information comprises finger-palm gesture information and wrist gesture information; and the normalized action information comprises normalized finger action information, normalized palm action information and normalized wrist action information.

[0084] The normalized action data information is calculated and processed to obtain the hand gesture information, comprising:

[0085] For any normalized action information, the normalized finger action information and the normalized palm action information corresponding to the normalized action information are calculated and processed to obtain the finger-palm gesture information corresponding to the normalized action information.

[0086] The normalized palm action information and the normalized wrist action information corresponding to the normalized action information are calculated and processed to obtain the wrist gesture information corresponding to the normalized action information.

[0087] It should be noted that since the collected hand information is several, the normalized action information obtained by normalizing the hand action information is also several. The processing method of each normalized action information is consistent.

[0088] It can be seen that the gesture feature recognition method described in the embodiment of the application is beneficial to improve the gesture feature recognition efficiency and accuracy, thereby improving the online real-time performance of gesture feature recognition and expanding the gesture action recognition classification category.

[0089] In another optional embodiment, the normalized finger action information and the normalized palm action information corresponding to the normalized action information are calculated and processed to obtain the finger-palm gesture information corresponding to the normalized action information, comprising:

[0090] The normalized finger action information and the normalized palm action information corresponding to the normalized action information are calculated and processed by using a first pose calculation model to obtain spatial pose information;

[0091] The first pose calculation model is:

[0092] ;

[0093] In the formula, is a spatial pose vector corresponding to the spatial pose information; is a normalized palm action vector corresponding to the normalized palm action information; is a normalized finger action vector corresponding to the normalized finger action information;

[0094] The spatial pose information is coordinate-converted by using a first coordinate conversion model to obtain the finger-palm gesture information corresponding to the normalized action information;

[0095] The first coordinate conversion model is:

[0096] ;

[0097] In the formula, is a finger-palm gesture vector corresponding to the finger-palm gesture information; is a spatial pose vector.

[0098] It should be noted that the finger-palm gesture information includes finger gesture information and palm gesture information.

[0099] In this optional embodiment, as an optional implementation, it is judged whether the hand feature corresponding to the spatial pose information is a finger feature, to obtain a feature judgment result;

[0100] When the feature judgment result is yes, the second vector element in the finger-palm gesture vector is determined as the finger gesture information;

[0101] When the feature judgment result is no, the finger-palm gesture vector is determined as the palm gesture information.

[0102] It should be noted that the palm gesture information includes three palm gesture values, i.e., a pitch value, a yaw value and a roll value. Further, the pitch value, the yaw value and the roll value correspond to , and respectively.

[0103] It can be seen that the gesture feature recognition method described in the embodiments of the present application is beneficial to improve the gesture feature recognition efficiency and accuracy, and further improve the online real-time performance of gesture feature recognition, and expand the gesture action recognition classification category.

[0104] In yet another optional embodiment, the normalized palm action information and the normalized wrist action information corresponding to the normalized action information are calculated and processed to obtain the wrist pose information corresponding to the normalized action information, including:

[0105] Obtaining a rotation parameter;

[0106] Performing transformation calculation processing on the rotation parameter to obtain normalized coordinate transformation quaternion information;

[0107] Performing product calculation on the normalized coordinate transformation quaternion information and the normalized palm action information corresponding to the normalized action information to obtain palm coordinate transformation information;

[0108] Using a second pose calculation model to calculate and process the normalized wrist action information and the palm coordinate transformation information to obtain wrist pose information;

[0109] The second pose calculation model is:

[0110] ;

[0111] wherein, is a wrist pose vector corresponding to the wrist pose information; is a normalized wrist motion vector corresponding to the normalized wrist motion information; is a palm coordinate transformation vector corresponding to the palm coordinate transformation information; is an inverse matrix function;

[0112] The wrist pose information is subjected to coordinate transformation processing to obtain the wrist pose information corresponding to the normalized motion information.

[0113] It should be noted that the rotation parameters include a first rotation angle and a second rotation angle . Further, the first rotation angle is a rotation angle about the x-axis, and the second rotation angle is a rotation angle about the z-axis.

[0114] In this optional embodiment, as an optional implementation, the rotation parameters are obtained by:

[0115] determining whether the hand type corresponding to the hand acquisition information is a target type to obtain a type determination result;

[0116] when the type determination result is yes, the first rotation angle is determined to be 90° and the second rotation angle is determined to be 0°;

[0117] when the type determination result is no, the first rotation angle is determined to be -90° and the second rotation angle is determined to be 0°.

[0118] It should be noted that the hand type represents the position of the hand corresponding to the gesture acquired. Further, the target type includes a left hand or a right hand. The determination of the target type can be dynamically set or pre-set, and the embodiments of the present application are not limited.

[0119] It should be noted that the coordinate transformation processing of the wrist pose information is based on the first coordinate transformation model.

[0120] It can be seen that the gesture feature recognition method described in the embodiments of the present application is beneficial to improve the gesture feature recognition efficiency and accuracy, thereby improving the online real-time performance of gesture feature recognition and expanding the gesture motion recognition classification categories.

[0121] In an optional embodiment, the rotation parameters include a first rotation angle and a second rotation angle; the first rotation angle and the second rotation angle are rotation angles relative to two directions perpendicular to each other;

[0122] The rotation parameters are subjected to transformation calculation processing to obtain normalized coordinate transformation quaternion information, including:

[0123] The first rotation angle and the second rotation angle are respectively subjected to a sine-cosine calculation to obtain first angle value information and second angle value information;

[0124] The first angle value information and the second angle value information are respectively filled into a preset first angle vector and a second angle vector to obtain first target angle vector information and second target angle vector information;

[0125] The first target angle vector information and the second target angle vector information are subjected to a cross multiplication calculation to obtain coordinate conversion vector information;

[0126] The coordinate conversion vector is subjected to a normalization processing to obtain normalized coordinate conversion quaternion information.

[0127] In the optional embodiment, as an optional implementation, the above-mentioned sine-cosine calculation of the first rotation angle and the second rotation angle respectively to obtain the first angle value information and the second angle value information comprises:

[0128] The first rotation angle is calculated by using a first angle calculation model to obtain the first angle value information;

[0129] The first angle calculation model is:

[0130] ;

[0131] ;

[0132] In the formula, is a first angle value in the first angle value information; is a second angle value in the first angle value information;

[0133] The second rotation angle is calculated by using a second angle calculation model to obtain the second angle value information;

[0134] The second angle calculation model is:

[0135] ;

[0136] ;

[0137] In the formula, is a third angle value in the second angle value information; is a fourth angle value in the second angle value information.

[0138] It should be noted that the first angle vector and the second angle vector are and .

[0139] It should be noted that the cross product mentioned above is a directed multiplication of two vectors, such as multiplying any element of the first angle vector by all elements of the second angle vector in sequence to obtain a new coordinate transformation vector.

[0140] It should be noted that the above normalization process for the coordinate transformation vector is based on a normalization model.

[0141] It is evident that implementing the gesture feature recognition method described in the embodiments of the present invention is beneficial to improving the efficiency and accuracy of gesture feature recognition, thereby improving the online real-time performance of gesture feature recognition and expanding the classification categories of gesture action recognition.

[0142] In another optional embodiment, target gesture feature information is obtained based on hand pose information, including:

[0143] For any target attitude information, the data values ​​in the target attitude information are filled into a preset attitude vector to obtain the target attitude vector corresponding to the target attitude information; the target attitude vector includes 9 vector elements;

[0144] According to the data acquisition order of the hand motion information, all target posture vectors are arranged sequentially to obtain the target posture matrix;

[0145] Based on the target pose matrix, the target gesture feature information is determined.

[0146] It should be noted that the aforementioned posture vector can be [the flexion and extension of the little finger, the flexion and extension of the ring finger, the flexion and extension of the middle finger, the flexion and extension of the ring finger, the abduction and adduction of the thumb, the internal and external rotation of the thumb, the pitch value of the palm, the yaw value, and the roll value]. Furthermore, the flexion and extension of the little finger, the flexion and extension of the ring finger, the flexion and extension of the middle finger, and the flexion and extension of the ring finger are the numerical values ​​corresponding to the finger posture information. The abduction and adduction of the thumb and the internal and external rotation of the thumb are the numerical values ​​corresponding to the wrist posture information.

[0147] It should be noted that the above determination of target gesture feature information based on the target pose matrix involves associating the target pose matrix with hand identity information to obtain the target gesture feature information that is finally input into the neural network-based recognition model.

[0148] It should be noted that the above target attitude matrix can be... The matrix is ​​a set of vectors, where each row represents a pose vector. Furthermore, M is 9. N is determined based on a neural network recognition model.

[0149] It should be noted that the data collection order of the hand movement information above is based on the time sequence of collection.

[0150] It is evident that implementing the gesture feature recognition method described in the embodiments of the present invention is beneficial to improving the efficiency and accuracy of gesture feature recognition, thereby improving the online real-time performance of gesture feature recognition and expanding the classification categories of gesture action recognition.

[0151] Example 2

[0152] Please see Figure 2 , Figure 2 This is a schematic diagram of the structure of a gesture feature recognition device disclosed in an embodiment of the present invention. Figure 2 The described device can be applied in warehouse management systems, such as local servers or cloud servers for warehouse logistics gesture recognition management, etc., and the embodiments of the present invention are not limited thereto. Figure 2 As shown, the device may include:

[0153] The acquisition module 201 is used to acquire hand movement information;

[0154] The processing module 202 is used to calculate and process hand motion information to obtain hand posture information; the hand posture information includes several target posture information; the target posture information represents the motion characteristics of the fingers and palm;

[0155] The determination module 203 is used to determine the target gesture feature information based on the hand posture information.

[0156] It is evident that implementation Figure 2 The described gesture feature recognition device is beneficial for improving the efficiency and accuracy of gesture feature recognition, thereby improving the online real-time performance of gesture feature recognition and expanding the categories of gesture action recognition.

[0157] In another alternative embodiment, such as Figure 2 As shown, the processing module 202 calculates and processes the hand motion information to obtain hand posture information, including:

[0158] The normalization model is used to normalize the hand motion information to obtain normalized motion data information; the normalized motion data information includes several normalized motion information.

[0159] The normalization model is as follows:

[0160] ;

[0161] In the formula, This is the normalized action vector corresponding to the normalized action information; This refers to the original hand data vector corresponding to the original hand data information in the hand movement information; , , and The first hand original value, the second hand original value, the third hand original value and the fourth hand original value are sequentially distributed in the hand original data vector respectively.

[0162] The normalized action data information is calculated and processed to obtain the hand posture information.

[0163] It can be seen that the embodiment Figure 2 The gesture feature recognition device described is beneficial to improve the gesture feature recognition efficiency and precision, and further improve the gesture feature recognition online real-time performance and expand the gesture action recognition classification category.

[0164] In another optional embodiment, as Figure 2 shown, the target posture information includes the finger-palm posture information and the wrist posture information; and the normalized action information includes the normalized finger action information, the normalized palm action information and the normalized wrist action information.

[0165] The processing module 202 calculates and processes the normalized action data information to obtain the hand posture information, including:

[0166] For any normalized action information, the normalized finger action information and the normalized palm action information corresponding to the normalized action information are calculated and processed to obtain the finger-palm posture information corresponding to the normalized action information.

[0167] The normalized palm action information and the normalized wrist action information corresponding to the normalized action information are calculated and processed to obtain the wrist posture information corresponding to the normalized action information.

[0168] It can be seen that the embodiment Figure 2 The gesture feature recognition device described is beneficial to improve the gesture feature recognition efficiency and precision, and further improve the gesture feature recognition online real-time performance and expand the gesture action recognition classification category.

[0169] In another optional embodiment, as Figure 2 shown, the processing module 202 calculates and processes the normalized finger action information and the normalized palm action information corresponding to the normalized action information to obtain the finger-palm posture information corresponding to the normalized action information, including:

[0170] The normalized finger action information and the normalized palm action information corresponding to the normalized action information are calculated and processed by using the first pose calculation model to obtain the spatial pose information.

[0171] The first pose calculation model is:

[0172] ;

[0173] In the formula, the spatial pose information corresponds to a spatial pose vector. ​The normalized palm movement vector corresponding to the normalized palm movement information; The normalized finger motion vector corresponding to the normalized finger motion information;

[0174] The spatial pose information is transformed using the first coordinate transformation model to obtain the finger and palm pose information corresponding to the normalized motion information.

[0175] The first coordinate transformation model is as follows:

[0176] ;

[0177] In the formula, This is the finger and palm posture vector corresponding to the finger and palm posture information; This is the spatial pose vector.

[0178] It is evident that implementation Figure 2 The described gesture feature recognition device is beneficial to improving the efficiency and accuracy of gesture feature recognition, thereby improving the online real-time performance of gesture feature recognition and expanding the categories of gesture action recognition.

[0179] In yet another alternative embodiment, such as Figure 2 As shown, the processing module 202 calculates and processes the normalized palm movement information and normalized wrist movement information corresponding to the normalized movement information to obtain the wrist posture information corresponding to the normalized movement information, including:

[0180] Obtain rotation parameters;

[0181] The rotation parameters are transformed and calculated to obtain the normalized coordinate transformation quaternary information;

[0182] The palm coordinate transformation information is obtained by multiplying the normalized coordinate transformation quaternary information and the normalized palm action information corresponding to the normalized action information.

[0183] The wrist posture information is obtained by calculating and processing the normalized wrist motion information and palm coordinate transformation information using the second pose calculation model.

[0184] The second pose calculation model is as follows:

[0185] ;

[0186] In the formula, This is the wrist pose vector corresponding to the wrist pose information. The normalized wrist motion vector corresponding to the normalized wrist motion information; This is the palm coordinate transformation vector corresponding to the palm coordinate transformation information; To find the inverse matrix function;

[0187] The wrist posture information is processed by coordinate transformation to obtain the wrist posture information corresponding to the normalized action information.

[0188] It can be seen that the embodiment Figure 2 The gesture feature recognition device described is advantageous in improving gesture feature recognition efficiency and precision, thereby improving gesture feature recognition online real-time performance and expanding gesture action recognition classification categories.

[0189] In yet another optional embodiment, as Figure 2 shown, the rotation parameters include a first rotation angle and a second rotation angle; the first rotation angle and the second rotation angle are rotation angles relative to two directions perpendicular to each other;

[0190] The processing module 202 performs transformation and calculation processing on the rotation parameters to obtain normalized coordinate transformation quaternion information, including:

[0191] The first rotation angle and the second rotation angle are respectively subjected to sine and cosine calculation to obtain first angle value information and second angle value information;

[0192] The first angle value information and the second angle value information are respectively filled into a preset first angle vector and a second angle vector to obtain first target angle vector information and second target angle vector information;

[0193] The first target angle vector information and the second target angle vector information are subjected to cross multiplication calculation processing to obtain coordinate transformation vector information;

[0194] The coordinate transformation vector is subjected to normalization processing to obtain normalized coordinate transformation quaternion information.

[0195] It can be seen that the embodiment Figure 2 The gesture feature recognition device described is advantageous in improving gesture feature recognition efficiency and precision, thereby improving gesture feature recognition online real-time performance and expanding gesture action recognition classification categories.

[0196] In yet another optional embodiment, as Figure 3 shown, the determination module 203 obtains target gesture feature information based on the hand posture information, including:

[0197] For any target posture information, data values in the target posture information are filled into a preset posture vector to obtain a target posture vector corresponding to the target posture information; the target posture vector includes nine vector elements;

[0198] All target posture vectors are sequentially arranged according to a data collection sequence of hand collection information in the hand action information to obtain a target posture matrix;

[0199] The target gesture feature information is determined based on the target posture matrix.

[0200] It can be seen that the embodiment Figure 3 The gesture feature recognition device described has the advantages of improving the gesture feature recognition efficiency and precision, and further improving the online real-time performance of gesture feature recognition and expanding the gesture action recognition classification category.

[0201] Embodiment three

[0202] Please refer to Figure 3 , Figure 3 is another structural schematic diagram of a gesture feature recognition device disclosed by the embodiment of the application. Among them, ​ The device described can be applied to a warehouse management system, such as a local server or a cloud server for warehouse logistics gesture feature recognition management, and the embodiment of the application is not limited. As shown in the figure, ​ The device can include:

[0203] The memory 301 stores executable program codes;

[0204] The processor 302 is coupled to the memory 301;

[0205] The processor 302 calls the executable program codes stored in the memory 301, and is used to execute the steps in the gesture feature recognition method described in embodiment one.

[0206] Embodiment four

[0207] The embodiment of the application discloses a computer readable storage medium which stores a computer program for electronic data exchange, wherein the computer program causes a computer to execute the steps in the gesture feature recognition method described in embodiment one.

[0208] Embodiment five

[0209] The embodiment of the application discloses a computer program product, which includes a non-transitory computer readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute the steps in the gesture feature recognition method described in embodiment one.

[0210] The device embodiments described above are only schematic, wherein the modules illustrated as separate components can or can not be physically separated, and the components illustrated as modules can or can not be physical modules, that is, they can be located in one place, or distributed on multiple network modules. According to actual needs, some or all of the modules can be selected to achieve the purpose of the embodiment scheme. Those skilled in the art can understand and implement without creative labor.

[0211] Those skilled in the art can clearly understand the implementation of the various embodiments by means of software and necessary general hardware platforms through the above specific description of the embodiments, and of course, the embodiments can also be implemented by hardware. Based on such understanding, the above technical solutions, essentially or in other words, the part of the prior art that contributes to the present application can be embodied in the form of a software product. The computer software product can be stored in a computer readable storage medium, including a Read-Only Memory (ROM), a Random Access Memory (RAM), a Programmable Read-only Memory (PROM), an Erasable Programmable Read Only Memory (EPROM), a One-time Programmable Read-Only Memory (OTPROM), an Electrically-Erasable Programmable Read-Only Memory (EEPROM), a Compact Disc Read-Only Memory (CD-ROM) or other optical disk storage, a magnetic disk storage, a magnetic tape storage, or any other medium that can be used to carry or store data in a computer readable manner.

[0212] Finally, it should be noted that: the gesture feature recognition method and device disclosed by the embodiments of the present application are only the preferred embodiments of the present application, and are used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: the technical solutions recorded in the foregoing embodiments can still be modified, or some technical features can be replaced by equivalents; and these modifications or replacements do not make the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method of gesture feature recognition, characterized in that, The method comprises: acquiring hand action information; performing calculation processing on the hand action information to obtain hand posture information; the hand posture information comprises a plurality of target posture information; the target posture information represents the action feature of fingers and palms; based on the hand posture information, target hand gesture feature information is determined; wherein, the calculation processing on the hand action information to obtain hand posture information comprises: performing normalization calculation processing on the hand action information by using a normalization model to obtain normalized action data information; the normalized action data information comprises a plurality of normalized action information; wherein, the normalization model is: ； In the formula, is a normalized action vector corresponding to the normalized action information; is a hand original data vector corresponding to the hand original data information in the hand action information; , , and are a first hand original value, a second hand original value, a third hand original value and a fourth hand original value distributed in the hand original data vector in sequence respectively. performing calculation processing on the normalized action data information to obtain hand posture information; wherein, the target posture information comprises finger-palm posture information and wrist posture information; the normalized action information comprises normalized finger action information, normalized palm action information and normalized wrist action information; the calculation processing on the normalized action data information to obtain hand posture information comprises: for any normalized action information, performing calculation processing on the normalized finger action information and the normalized palm action information corresponding to the normalized action information to obtain the finger-palm posture information corresponding to the normalized action information; performing calculation processing on the normalized palm action information and the normalized wrist action information corresponding to the normalized action information to obtain the wrist posture information corresponding to the normalized action information.

2. The gesture feature recognition method of claim 1, wherein, the calculation processing on the normalized finger action information and the normalized palm action information corresponding to the normalized action information to obtain the finger-palm posture information corresponding to the normalized action information comprises: performing calculation processing on the normalized finger action information and the normalized palm action information corresponding to the normalized action information by using a first pose calculation model to obtain spatial pose information; wherein, the first pose calculation model is: ； In the formula, is a spatial pose vector corresponding to the spatial pose information; is a normalized palm action vector corresponding to the normalized palm action information; is the normalized finger action vector corresponding to the normalized finger action information; performing coordinate conversion processing on the spatial pose information by using a first coordinate conversion model to obtain the finger-palm posture information corresponding to the normalized action information; wherein, the first coordinate conversion model is: ； In the formula, is a finger-palm posture vector corresponding to the finger-palm posture information; is a spatial pose vector.

3. The gesture feature recognition method of claim 2, wherein, the calculation processing on the normalized palm action information and the normalized wrist action information corresponding to the normalized action information to obtain the wrist posture information corresponding to the normalized action information comprises: obtaining a rotation parameter; performing transformation calculation processing on the rotation parameter to obtain normalized coordinate transformation quaternion information; performing product calculation on the normalized coordinate transformation quaternion information and the normalized palm action information corresponding to the normalized action information to obtain palm coordinate transformation information; performing calculation processing on the normalized wrist action information and the palm coordinate transformation information by using a second pose calculation model to obtain wrist posture information; wherein, the second pose calculation model is: ； In the formula, is a wrist pose vector corresponding to the wrist pose information; is a normalized wrist motion vector corresponding to the normalized wrist motion information; is a palm coordinate transformation vector corresponding to the palm coordinate transformation information; is an inverse matrix function; performing coordinate conversion processing on the wrist posture information to obtain the wrist posture information corresponding to the normalized action information.

4. The gesture feature recognition method of claim 3, wherein, the rotation parameter comprises a first rotation angle and a second rotation angle; the first rotation angle and the second rotation angle are rotation angles relative to two directions perpendicular to each other; The rotation parameter conversion calculation processing obtains normalized coordinate conversion quaternion information, including: Sine and cosine calculation is performed on the first rotation angle and the second rotation angle respectively to obtain first angle value information and second angle value information; The first angle value information and the second angle value information are filled into a preset first angle vector and a second angle vector respectively to obtain first target angle vector information and second target angle vector information; Cross multiplication calculation processing is performed on the first target angle vector information and the second target angle vector information to obtain coordinate conversion vector information; Normalization processing is performed on the coordinate conversion vector to obtain normalized coordinate conversion quaternion information.

5. The gesture feature recognition method of claim 1, wherein, The target gesture feature information is obtained based on the hand posture information, including: For any target posture information, data values in the target posture information are filled into a preset posture vector to obtain a target posture vector corresponding to the target posture information; the target posture vector includes nine vector elements; All target posture vectors are sequentially arranged according to the data acquisition sequence of the hand acquisition information in the hand action information to obtain a target posture matrix; The target gesture feature information is determined based on the target posture matrix.

6. A gesture feature recognition apparatus, characterized by, The device includes: An acquisition module for acquiring hand action information; A processing module for performing calculation processing on the hand action information to obtain hand posture information; the hand posture information includes a plurality of target posture information; the target posture information represents the action feature of the fingers and the palm; A determination module for determining target gesture feature information based on the hand posture information; The calculation processing on the hand action information to obtain hand posture information includes: Normalization calculation processing is performed on the hand action information by using a normalization model to obtain normalized action data information; the normalized action data information includes a plurality of normalized action information; The normalization model is: ； In the formula, is a normalized action vector corresponding to the normalized action information; is a hand original data vector corresponding to the hand original data information in the hand action information; , , and are a first hand original value, a second hand original value, a third hand original value and a fourth hand original value distributed in the hand original data vector in sequence respectively. The normalized action data information is processed to obtain hand posture information; The target posture information includes finger-palm posture information and wrist posture information; the normalized action information includes normalized finger action information, normalized palm action information, and normalized wrist action information; The calculation processing on the normalized action data information to obtain hand posture information includes: For any normalized action information, the normalized finger action information and the normalized palm action information corresponding to the normalized action information are processed to obtain the finger-palm posture information corresponding to the normalized action information; The normalized palm action information and the normalized wrist action information corresponding to the normalized action information are processed to obtain the wrist posture information corresponding to the normalized action information.

7. A gesture feature recognition apparatus, characterized by, The device includes: A memory storing executable program code; A processor coupled with the memory; The processor invokes the executable program code stored in the memory to execute the gesture feature recognition method according to any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer instructions, which are invoked to perform the gesture feature recognition method according to any one of claims 1-5.

Images