A multifunctional grinding machine control method and system based on gesture recognition

Abstract

The present invention relates to the field of artificial intelligence technology, and specifically relates to a method and system for controlling a multifunctional grinding machine based on gesture recognition. The method comprises: determining whether the acceleration change sequence of the multifunctional grinding machine is within a working floating range; if so, performing a palm object recognition operation on surrounding environmental data and a filtering operation based on the user's free hand to obtain a user's free hand object; identifying the gesture of the user's free hand object to obtain an action gesture; if not, performing motion trajectory recognition on the acceleration change sequence to obtain a motion trajectory; and upon detecting the action gesture or motion trajectory, utilizing a pre-built association database for querying gesture motion-service functions to perform a function query on the action gesture or motion trajectory to obtain a target function, and then executing the target function using the multifunctional grinding machine. The present invention can improve the efficiency of grinding machine function switching.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a multifunctional grinder control method and system based on gesture recognition. Background Art

[0002] A grinder is a simple electric tool that is used in various fields of daily life. For example, large equipment can be used to grind building materials, and small equipment can be used to carve jade and nails.

[0003] Today's grinding machines incorporate multiple functions, such as forward and reverse rotation and various levels of force, to accommodate diverse work needs. These grinding functions are often switched via buttons. The grinding process generates a large amount of powder, which hinders cleaning dirt and bacteria from the gaps between buttons. Furthermore, the increasing number of buttons often requires users to interrupt their work flow to switch functions, impacting efficiency. Summary of the Invention

[0004] The present invention provides a multifunctional grinding machine control method based on gesture recognition, the main purpose of which is to improve the efficiency of grinding machine function switching.

[0005] To achieve the above objectives, the present invention provides a multifunctional grinding machine control method based on gesture recognition, comprising:

[0006] Obtain a multifunctional grinder, wherein the multifunctional grinder includes an acceleration sensor and an environmental sensing device;

[0007] Using the acceleration sensor, the acceleration of the multifunctional grinder is monitored to obtain an acceleration change sequence;

[0008] Determining whether the acceleration change sequence is within a preset working floating range;

[0009] When it is determined that the acceleration change sequence is within the working floating range, using the environmental sensing device to obtain ambient environment data of the multifunctional grinder, performing a palm object recognition operation on the ambient environment data to obtain a palm object set, and performing a screening operation on the palm object set based on the user's free hand to obtain a user's free hand object;

[0010] Performing key node monitoring on the user's idle hand object to obtain a sequence of changes in the user's palm skeleton, and performing gesture recognition on the sequence of changes in the user's palm skeleton based on a pre-built gesture database to obtain an action gesture;

[0011] When it is determined that the acceleration change sequence is not within the working floating range, performing motion trajectory recognition on the acceleration change sequence to obtain a motion trajectory;

[0012] determining whether the action gesture or the motion trajectory is detected;

[0013] When the action gesture or the motion trajectory is not detected, returning to the above step of using the pre-built acceleration sensor to monitor the acceleration of the multifunctional grinder to obtain an acceleration change sequence;

[0014] When the action gesture or the motion trajectory is detected, a pre-built association database for querying gesture motion-service functions is used to perform a function query on the action gesture or motion trajectory to obtain a target function, and the multi-functional grinder is used to execute the target function.

[0015] Optionally, performing a palm object recognition operation on the surrounding environment data to obtain a palm object set includes:

[0016] Performing Gaussian filtering on the surrounding environment data to obtain noise-reduced environmental data;

[0017] Performing an edge recognition operation on the noise reduction environment data to obtain an object set;

[0018] A preset palm type classification and judgment operation is performed on each object in the object set to obtain a palm set.

[0019] Optionally, performing a screening operation on the palm object set based on the user's free hand to obtain the user's free hand object includes:

[0020] performing palm gesture recognition on each palm object in the palm object set to obtain a palm gesture feature set;

[0021] According to the palm gesture feature set, a type recognition operation is performed on the palm object set to obtain a grinder execution hand and a customer target hand;

[0022] The palm objects other than the grinder execution hand and the customer target hand in the palm object set are determined as user idle hand objects.

[0023] Optionally, the determining whether the action gesture or the motion trajectory is detected includes:

[0024] Get an action gesture, where the action gesture is represented by:

[0025] v=F[Δx1,Δy1,Δz1,…,Δx k , Δy k , Δz k ]

[0026] Where v represents the action gesture, F[·] represents gesture recognition, Δx1, Δy1, Δz1 represent the palm information changes of the first element in the user's palm skeleton change sequence in the three dimensions of x, y, and z, Δx k ,Δy k ,Δz k Indicates the palm information change of the kth element in the user's palm skeleton change sequence in the x, y, and z dimensions;

[0027] Obtain a motion trajectory, wherein the motion trajectory is expressed as:

[0028]

[0029] Where PSD(f) represents the motion trajectory, is the energy spectrum density in the frequency domain signal of the acceleration change sequence, f represents the frequency, a(t) represents the acceleration change sequence, e -j2πft / N represents the complex exponential function, j represents the imaginary unit, t represents the time, and N represents the number of sampling points;

[0030] The action gesture and motion trajectory are subjected to feature splicing to obtain a joint feature vector of both hands, wherein the joint feature vector of both hands is expressed as:

[0031] G=[v,log(PSD(f dominant ))]

[0032] Where G represents the joint feature vector of both hands, f dominant The main signal in the frequency domain signal representing the acceleration change sequence;

[0033] A binary classification judgment based on the action gesture or the motion trajectory is performed according to the joint feature vector of the two hands.

[0034] Optionally, monitoring key nodes of the user's idle hand object to obtain a user palm skeleton change sequence includes:

[0035] Performing a feature extraction operation on the user's idle hand object to obtain a palm feature distribution set, and performing a feature weight sorting operation based on principal component feature analysis on the palm feature distribution set to obtain a palm key feature set;

[0036] In the user's idle hand object, performing position coordinate selection on each palm key feature in the palm key feature set to obtain a palm key node set;

[0037] Performing palm skeleton connection on the palm key node set to obtain a palm point-line skeleton graph;

[0038] The change path of the palm point-line skeleton diagram within a preset time period is recorded to obtain a user palm skeleton change sequence.

[0039] Optionally, performing gesture recognition on the user's palm skeleton change sequence based on a pre-built gesture database to obtain an action gesture includes:

[0040] According to a preset amplitude threshold, a filtering operation based on the movement amplitude is performed on the user's palm skeleton change sequence to obtain a set of fingers with the largest amplitude movement;

[0041] acquiring, from the user's palm skeleton change sequence, finger movements of each finger with the largest amplitude movement in the set of fingers with the largest amplitude movement, to obtain a finger skeleton change set;

[0042] performing recognition operations based on single finger movements and coordinated finger movements on the finger skeleton change set to obtain a finger movement set;

[0043] According to the pre-built gesture database, a similarity clustering and screening operation is performed on the finger action set to obtain action gestures.

[0044] Optionally, performing motion trajectory recognition on the acceleration change sequence to obtain the motion trajectory includes:

[0045] Extracting waveform features of the acceleration change sequence to obtain waveform amplitude information and waveform duration;

[0046] Performing a Fourier transform operation on the acceleration change sequence to obtain frequency domain signal features, and performing frequency feature extraction on the frequency domain signal features to obtain main frequency feature information and period feature information;

[0047] Key motion trajectory identification is performed on the waveform amplitude information, waveform duration, main frequency characteristic information and period characteristic information to obtain a motion trajectory.

[0048] Optionally, before using a pre-built database for querying gesture motion-service function association, the method further includes:

[0049] When starting the pre-built action entry service, obtaining the user's repeatedly executed actions on the multi-functional grinder;

[0050] Fitting the repeatedly executed action to obtain an input action;

[0051] Obtain the action name configured by the user, obtain the function link configured by the user, and construct a key-value pair relationship between the function link and the input action;

[0052] According to the action name, the key-value pair relationship is entered into a pre-built association database for querying gesture motion-service function.

[0053] Optionally, after determining whether the action gesture or the motion trajectory is detected, the method further includes:

[0054] Obtain the successfully detected action gestures or motion trajectories within a preset time period to obtain historical execution actions;

[0055] Identifying a change pattern of action execution in the historical execution actions, and performing fuzzy processing based on action amplitude on the input action in the key-value pair relationship according to the change pattern of action execution to obtain an updated input action;

[0056] According to the update entry action, the key-value pair relationship in the association database for querying gesture motion-service function is updated.

[0057] To achieve the above object, the present invention further provides a multifunctional grinding machine control system based on gesture recognition, comprising:

[0058] A grinder acquisition module, configured to acquire a multifunctional grinder, wherein the multifunctional grinder includes an acceleration sensor and an environmental sensing device;

[0059] An information acquisition module, configured to monitor the acceleration of the multifunctional grinder using the acceleration sensor to obtain an acceleration change sequence;

[0060] a user gesture recognition module, configured to determine whether the acceleration change sequence is within a preset working floating range, and when it is determined that the acceleration change sequence is within the working floating range, using the environmental sensing device to obtain ambient environment data of the multifunctional grinder, performing a palm object recognition operation on the ambient environment data to obtain a palm object set, performing a screening operation on the palm object set based on the user's idle hand to obtain a user's idle hand object, and performing key node monitoring on the user's idle hand object to obtain a user's palm skeleton change sequence, and performing gesture recognition on the user's palm skeleton change sequence based on a pre-built gesture database to obtain an action gesture;

[0061] a grinding machine trajectory recognition module, configured to, when determining that the acceleration change sequence is not within the working floating range, perform motion trajectory recognition on the acceleration change sequence to obtain a motion trajectory;

[0062] The instruction control module is used to, when the action gesture or the motion trajectory is not detected, return to the above-mentioned step of using the pre-built acceleration sensor to monitor the acceleration of the multi-functional grinder to obtain an acceleration change sequence; and when the action gesture or the motion trajectory is detected, use a pre-built association database for querying gesture motion-service functions to perform a function query on the action gesture or motion trajectory to obtain a target function, and use the multi-functional grinder to execute the target function.

[0063] In order to solve the above problem, the present invention further provides an electronic device, comprising:

[0064] a memory storing at least one instruction;

[0065] The processor executes the instructions stored in the memory to implement the above-mentioned multifunctional grinder control method based on gesture recognition.

[0066] In order to solve the above problems, the present invention also provides a computer-readable storage medium, which stores at least one instruction. The at least one instruction is executed by a processor in an electronic device to implement the above-mentioned multi-functional grinder control method based on gesture recognition.

[0067] To address the issues described in the background art, the present invention first transforms a multifunctional grinder by replacing traditional buttons with gesture recognition. To improve the user's work continuity and efficiency, the present invention employs dual recognition of both the operator hand and the idle hand. The operator hand refers to the hand holding the multifunctional grinder, while the idle hand refers to the hand holding the customer's finger. Dual recognition involves functionally configuring the idle hand's gestures and the operator hand's motion trajectory. Both dual recognition operations utilize a combination of sensors and artificial intelligence, significantly improving detection accuracy and thus increasing the flexibility of the grinder's functional changes. Therefore, the present invention can improve the efficiency of grinding machine function switching. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 A schematic flow chart of a method for controlling a multifunctional grinding machine based on gesture recognition according to an embodiment of the present invention;

[0069] Figure 2 A functional module diagram of a multifunctional grinding machine control system based on gesture recognition provided by one embodiment of the present invention;

[0070] Figure 3 A schematic structural diagram of an electronic device for implementing the multifunctional grinder control method based on gesture recognition provided by an embodiment of the present invention.

[0071] Description of reference numerals:

[0072] 1. Electronic device; 10. Processor; 11. Memory; 12. Bus.

[0073] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0074] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0075] The embodiments of the present application provide a method for controlling a multifunctional grinder based on gesture recognition. The execution entity of the method includes, but is not limited to, at least one of electronic devices such as a server and a terminal that can be configured to execute the method provided by the embodiments of the present application. In other words, the method for controlling a multifunctional grinder based on gesture recognition can be executed by software or hardware installed on a terminal device or a server device, where the software can be a blockchain platform. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.

[0076] Reference Figure 1 FIG. 1 is a flow chart of a method for controlling a multifunctional grinding machine based on gesture recognition according to an embodiment of the present invention. In this embodiment, the method for controlling a multifunctional grinding machine based on gesture recognition includes:

[0077] S1. Obtain a multifunctional grinder, wherein the multifunctional grinder includes an acceleration sensor and an environmental sensing device.

[0078] The multifunctional grinding machine is a grinding machine with multiple adjustment operations.

[0079] The acceleration sensor is a gyroscope device that can be used to record the motion trajectory of the multi-function grinder. For example, an acceleration sensor (G-sensor), a magnetic sensor (M-sensor), an orientation sensor (O-sensor), a gyroscope sensor (Gyro-sensor), a gravity sensor (GV-sensor), a linear acceleration sensor (LA-sensor), and a rotation vector sensor.

[0080] The environmental sensing device is a short-range infrared scanning device that can observe palm information within 10 cm around the multifunctional grinder.

[0081] Specifically, in an embodiment of the present invention, the adjustment operations in the multifunctional grinder have been transformed from the traditional multi-button adjustment method to one that utilizes built-in sensors and artificial intelligence (AI) for collaborative recognition. These adjustments include at least 0-40 power settings, grinding direction reversal, pause, and start operations. The sensors are configured as gyroscopes and infrared scanners, and the AI ​​is a computing module with edge computing capabilities.

[0082] In the embodiment of the present invention, by means of intelligent recognition by sensors, it is possible to avoid adding buttons, thereby improving the cleanability of the multifunctional grinder.

[0083] S2. Using the acceleration sensor, monitor the acceleration of the multifunctional grinder to obtain an acceleration change sequence.

[0084] The acceleration monitoring is the process of converting the scale and direction of the gyroscope into the magnitude and direction of acceleration in real time. The acceleration change sequence is a record of the magnitude and direction of acceleration within a preset time period t.

[0085] Specifically, in the embodiment of the present invention, the preset time t is set to 2 seconds taking into account the actual working scenarios of manicure and carving. Therefore, the acceleration sensor can be used to obtain the monitoring data of the acceleration sensor in real time, and the acquired monitoring data can be cached for a period of 2 seconds to obtain an acceleration change sequence.

[0086] S3. Determine whether the acceleration change sequence is within a preset working floating range.

[0087] The working floating range is a set of preset numerical ranges for limiting acceleration, which can be adjusted according to the working environment.

[0088] Specifically, in this embodiment of the present invention, considering the manicure and engraving process, the multifunctional grinder should maintain relative stability with the customer's hand or object being engraved. Therefore, by determining whether each acceleration data point in the acceleration variation sequence is within a preset operating fluctuation range, it is possible to determine whether a stage of the manicure or engraving process has been completed and whether a change in grinding force or direction is needed to proceed to the next step.

[0089] In the subsequent explanation of the present invention, the term "grinding machine" represents the multifunctional grinder, the term "carving object" represents the customer's nails or the carved object, the term "operating hand" represents the hand holding the grinder, and the term "free hand" represents the hand fixing the carving object.

[0090] When it is determined that the acceleration change sequence is in the working floating range, S4, using the environmental perception device to obtain the surrounding environment data of the multifunctional grinder, performing a palm object recognition operation on the surrounding environment data to obtain a palm object set, and performing a screening operation based on the user's idle hand on the palm object set to obtain a user's idle hand object.

[0091] Specifically, in an embodiment of the present invention, when it is determined that the acceleration change sequence is within the working floating range, it indicates that the user's hand operating the grinder is relatively stable. If the user wants to perform other operations such as fine-tuning or re-sculpting at this time, the user can use one or two fingers of the hand that fixes the sculpture to perform an action, thereby adjusting the grinder function without any change in the operating hand, thereby greatly improving the smoothness of the working process and improving work efficiency.

[0092] The surrounding environment data refers to the recognition results of various objects within 10 cm around the grinder, including object type, object posture, etc.

[0093] The palm object recognition operation refers to an operation of extracting a palm object from various objects around the grinder using a pre-trained artificial intelligence model, and the palm object set is a recognition result of the palm object recognition operation.

[0094] The user's free hand refers to the above-mentioned free hand, and the user's free hand object refers to the information collection of the identified free hand, such as posture, position, etc.

[0095] Specifically, in an embodiment of the present invention, performing a palm object recognition operation on the surrounding environment data to obtain a palm object set includes:

[0096] Performing Gaussian filtering on the surrounding environment data to obtain noise-reduced environmental data;

[0097] Performing an edge recognition operation on the noise reduction environment data to obtain an object set;

[0098] A preset palm type classification and judgment operation is performed on each object in the object set to obtain a palm set.

[0099] The Gaussian filtering process refers to a method of performing noise reduction processing on the surrounding environment data by using a Gaussian filtering algorithm. The noise reduction environment data is the result of the Gaussian filtering process on the surrounding environment data.

[0100] The edge recognition operation is an operation of separating each object from the environment based on the color change in the data. The object set is a set of objects recognized around the grinding machine.

[0101] The preset palm type is an object type parameter configured by the present invention that can represent a human palm.

[0102] The classification judgment operation refers to the operation of judging whether each object in the object set is of the palm type or not. The palm set refers to the set of palms around the grinder.

[0103] Specifically, in an embodiment of the present invention, each pixel in the ambient environment data is smoothed using a Gaussian filtering algorithm, thereby reducing the noise in the ambient environment data to obtain noise-reduced ambient environment data. The present invention then uses a pre-built Sobel operator to perform edge recognition operations, finding all objects within 10 cm of the grinder, and obtaining an object set, such as [scissors, first hand, carving knife, second hand, third hand, etc.]. Furthermore, the present invention extracts hand objects from the object set by palm type, ignoring other objects, to obtain a palm set, such as [first hand, second hand, third hand].

[0104] Specifically, in an embodiment of the present invention, the filtering operation based on the user's free hands on the palm object set to obtain the user's free hand object includes:

[0105] performing palm gesture recognition on each palm object in the palm object set to obtain a palm gesture feature set;

[0106] According to the palm gesture feature set, a type recognition operation is performed on the palm object set to obtain a grinder execution hand and a customer target hand;

[0107] The palm objects other than the grinder execution hand and the customer target hand in the palm object set are determined as user idle hand objects.

[0108] Palm gesture recognition involves a feature extraction process using convolution kernels for palm gesture information. This process, for example, involves convolution, pooling, and flattening operations within a CNN. This and subsequent features such as feature extraction and model prediction are not limited in this embodiment of the present invention; any common CNN or other network architecture can be used. The palm gesture feature set is the feature extraction results for each palm object in the palm object set.

[0109] The type recognition operation refers to the operation of identifying whose hand the palm object belongs to. The polishing machine execution hand refers to the polishing machine working object (manicure hand), the polishing machine operating hand (user's hand) and the other free hand (the hand used to hold the manicure or carving object).

[0110] Specifically, in this embodiment of the present invention, by invoking a pre-trained neural network model, it is possible to extract the hand's posture, obtain a set of hand posture features, and further identify the owner of each hand, classifying it into a grinder's execution hand and a customer's target hand. By default, the present invention identifies all palm objects in the palm object set other than the grinder's execution hand and the customer's target hand as the user's free hand objects.

[0111] Specifically, in an embodiment of the present invention, before performing key node monitoring on the user's free hand object, the method further includes:

[0112] Determine whether the user's free hand object is an empty set;

[0113] When the user's free hand object is an empty set, determining that the multifunctional grinder does not use an action gesture as an instruction for switching a service function;

[0114] When the user's free hand object is not an empty set, the operation of performing key node monitoring on the user's free hand object continues.

[0115] The empty set is a set that does not contain any palm objects.

[0116] Specifically, in this embodiment of the present invention, since traditional work processes sometimes do not require the free hand to participate in the work, the user's free hand object may be an empty set. When the user's free hand object is an empty set, the present invention does not use the free hand's gestures as instructions for switching service functions. In other words, if the user's other hand is more than 10 cm away from the grinder, the present invention only uses the motion trajectory of the operator's hand using the grinder to adjust the function, such as "V"-shaped shaking, clockwise or counterclockwise circles, etc.

[0117] S5. Monitor key nodes of the user's idle hand object to obtain a user's palm skeleton change sequence, and perform gesture recognition on the user's palm skeleton change sequence based on a pre-built gesture database to obtain an action gesture.

[0118] Key node monitoring refers to the operation of monitoring the positions of points in the user's free hand object that have high palm feature weights. These points are locations corresponding to features that are important for palm recognition, such as fingers and knuckles.

[0119] The gesture database refers to a database storing preset gestures.

[0120] The gesture recognition operation refers to determining whether a gesture formed by a sequence of changes in the user's palm skeleton is in the gesture database. The action gesture is a gesture formed by a sequence of changes in the user's palm skeleton and existing in the gesture database.

[0121] Specifically, in an embodiment of the present invention, monitoring key nodes of the user's idle hand object to obtain a user's palm skeleton change sequence includes:

[0122] Performing a feature extraction operation on the user's idle hand object to obtain a palm feature distribution set, and performing a feature weight sorting operation based on principal component feature analysis on the palm feature distribution set to obtain a palm key feature set;

[0123] In the user's idle hand object, performing position coordinate selection on each palm key feature in the palm key feature set to obtain a palm key node set;

[0124] Performing palm skeleton connection on the palm key node set to obtain a palm point-line skeleton graph;

[0125] The change path of the palm point-line skeleton diagram within a preset time period is recorded to obtain a user palm skeleton change sequence.

[0126] The feature extraction operation is the same as the feature extraction process in S4 above, except that the field of view (or resolution) of feature extraction is adapted to extract each region of the palm. The palm feature distribution set is a collection of position information of each finger and finger joint on the user's idle hand object.

[0127] Principal component feature analysis (PCF) is an algorithm that describes the importance of each palm feature in a set of palm feature distributions to the palm object recognition result. For example, if a one-square-centimeter image of the back of the hand is used, the probability of predicting the object as a palm is 30%, while a one-square-centimeter image of the fingertip is used, the probability of predicting the object as a palm is 80%. Therefore, the feature weight of the fingertip feature is greater than that of the back of the hand feature.

[0128] The feature weight sorting operation refers to the process of sorting the results of the principal component feature analysis from large to small weights. The palm key feature set refers to the first 10 feature points in the result of the feature weight sorting operation, such as fingertips, joints, etc.

[0129] The position coordinate box selection refers to the process of marking various areas on the user's idle hand object using image processing tools. For example, each fingertip and knuckle on the user's idle hand object is marked with a 0.1 square centimeter box. The palm key node set represents the set of coordinates of each area selected on the user's idle hand object.

[0130] The palm skeleton connection refers to the operation of connecting the key points of the palm according to the shape characteristics of the palm. The palm point-line skeleton diagram is the result of connecting the key points of the palm.

[0131] The change path is the result of connecting the palm point-line skeleton diagrams at each time point in chronological order. For example, the index fingertip in the palm point-line skeleton diagram at the first time point is connected to the index fingertip in the palm point-line skeleton diagram at the second time point to form the change path of the index fingertip. The user palm skeleton change sequence is the result of recording the change paths of each palm key point in the user's palm skeleton.

[0132] In the embodiment of the present invention, the user's palm skeleton change sequence can be obtained according to the above description, and the user's palm skeleton change sequence can provide data support for subsequent recognition of gestures.

[0133] Specifically, in an embodiment of the present invention, performing gesture recognition on the user's palm skeleton change sequence based on a pre-built gesture database to obtain an action gesture includes:

[0134] According to a preset amplitude threshold, a filtering operation based on the movement amplitude is performed on the user's palm skeleton change sequence to obtain a set of fingers with the largest amplitude movement;

[0135] acquiring, from the user's palm skeleton change sequence, finger movements of each finger with the largest amplitude movement in the set of fingers with the largest amplitude movement, to obtain a finger skeleton change set;

[0136] performing recognition operations based on single finger movements and coordinated finger movements on the finger skeleton change set to obtain a finger movement set;

[0137] According to the pre-built gesture database, a similarity clustering and screening operation is performed on the finger action set to obtain action gestures.

[0138] The amplitude threshold is a parameter configured in the present invention and can be adjusted based on actual working scenarios. In the embodiments of the present invention, when one finger on the palm performs an action, other fingers will also produce some changes in conjunction. Therefore, the amplitude threshold can be configured to distinguish the amplitude of the finger that performs the main action from the amplitude of other fingers that passively change.

[0139] The screening operation refers to the operation of finding fingers whose movement amplitude is greater than the amplitude threshold. The maximum amplitude movement finger set is the result after screening, which may be one finger or multiple fingers.

[0140] The finger skeleton change set represents the motion information of each finger in the maximum amplitude motion finger set, which may be motion information of only one finger or information of multiple fingers performing movements in coordination.

[0141] The recognition operation based on single finger movements and coordinated finger movements is a process of integrating and quantifying finger movements through artificial intelligence. The finger movement set represents a set of individual finger movements. For example, a string of codes can be used to represent a particular finger.

[0142] The similarity clustering screening operation refers to calculating whether the finger movement is consistent with the pre-stored finger movement through a similarity algorithm, thereby increasing the probability of successful recognition.

[0143] Specifically, in an embodiment of the present invention, the fingers with the largest movement amplitudes are first screened based on the amplitude threshold to obtain the maximum amplitude movement finger set. Then, based on the maximum amplitude movement finger set and the palm skeleton change sequence, the specific movement information of the fingers with the largest movement amplitudes is obtained to obtain the finger skeleton change set. Then, the finger skeleton change set is subjected to movement recognition through a pre-built movement recognition neural network model to obtain a finger movement set. However, the movements in the finger movement set may all be useless movements, and whether to adjust the grinder function still requires checking the gesture database. If the corresponding finger movement is matched in the gesture database, it means that the user's finger movement is valid.

[0144] When it is determined that the acceleration change sequence is not in the working floating range, S6 , performing motion trajectory recognition on the acceleration change sequence to obtain a motion trajectory.

[0145] In this embodiment of the present invention, if the acceleration variation sequence is determined to be outside the operating range, it indicates that the operator has made some movement, causing the grinder to move drastically. At this point, the grinder may have already moved away from the idle hand. Therefore, the grinder's function switching can be achieved by simply considering whether the grinder's motion trajectory exhibits a specific pattern.

[0146] The motion trajectory recognition refers to the operation of predicting the motion trajectory of the grinder based on the change of acceleration. The motion trajectory is the path of the grinder's motion changes within a preset time period.

[0147] In detail, in an embodiment of the present invention, performing motion trajectory recognition on the acceleration change sequence to obtain the motion trajectory includes:

[0148] Extracting waveform features of the acceleration change sequence to obtain waveform amplitude information and waveform duration;

[0149] Performing a Fourier transform operation on the acceleration change sequence to obtain frequency domain signal features, and performing frequency feature extraction on the frequency domain signal features to obtain main frequency feature information and period feature information;

[0150] Key motion trajectory identification is performed on the waveform amplitude information, waveform duration, main frequency characteristic information and period characteristic information to obtain a motion trajectory.

[0151] The waveform feature extraction refers to the operation of obtaining time domain information in the acceleration change sequence. The waveform amplitude information and waveform duration are two results of the waveform feature extraction operation.

[0152] The Fourier transform operation is an operation for converting a time domain signal into frequency domain information. The frequency domain signal feature is a Fourier transform result of the acceleration change sequence.

[0153] Among them, the frequency feature extraction refers to the operation of obtaining the frequency features in the frequency domain signal features. The main frequency feature information and the periodic feature information represent two results of the frequency feature extraction process. The main frequency feature information can stabilize the hand shaking during the movement and retain the characteristics of the main action (for example, when a person draws a circle for the first time, the result is not too round, and there is jitter in the resulting line segment). The periodic feature information can identify which actions are performed twice or more, such as a grinder moving counterclockwise for two circles.

[0154] The key motion trajectory identification refers to the operation of extracting the main motion trajectory in the acceleration change sequence. The motion trajectory is the recognition result of the key motion trajectory identification, which only contains the intermediate steps in the acceleration change sequence, and the starting step and the ending step will be ignored.

[0155] Specifically, in an embodiment of the present invention, considering the complexity of motion recognition, it is expected to perform time domain analysis and frequency domain analysis on the acceleration change sequence, and further verify the time domain analysis results through the frequency domain analysis results, so as to achieve accurate recognition of the motion trajectory.

[0156] Specifically, in the embodiment of the present invention, a time domain analysis is first performed, and waveform feature extraction is performed through a pre-built neural network to obtain waveform amplitude information and waveform duration. Then, the time domain signal (acceleration change sequence) is converted into a frequency domain signal through the Fourier transform formula to obtain frequency domain signal features. The frequency domain signal features are then processed through a neural network that can extract frequency features to obtain main frequency feature information and periodic feature information. Finally, the waveform amplitude information, waveform duration, main frequency feature information and periodic feature information are subjected to full connection-based key motion trajectory identification through a fully connected network to obtain a motion trajectory. Among them, waveform feature extraction, frequency feature extraction and fully connected layers can be implemented using commonly used neural network structures, such as neural network models based on CNN and Transformer.

[0157] S7: Determine whether the action gesture or the motion trajectory is detected.

[0158] Specifically, in an embodiment of the present invention, determining whether the action gesture or the motion trajectory is detected includes:

[0159] Get an action gesture, where the action gesture is represented by:

[0160] v=F[Δx1,Δy1,Δz1,…,Δx k , Δy k , Δz k ]

[0161] Where v represents the action gesture, F[·] represents gesture recognition, Δx1, Δy1, Δz1 represent the palm information changes in the x, y, and z directions of the first element in the palm skeleton change sequence, and Δx k ,Δy k ,Δz k Indicates the change in the palm information in the x, y, and z directions of the kth element in the user's palm skeleton change sequence;

[0162] Obtain a motion trajectory, wherein the motion trajectory is expressed as:

[0163]

[0164] Where PSD(f) represents the motion trajectory, is the energy spectrum density in the frequency domain signal of the acceleration change sequence, f represents the frequency, a(t) represents the acceleration change sequence, e -j2πft / N represents the complex exponential function, j represents the imaginary unit, t represents the time, and N represents the number of sampling points;

[0165] The action gesture and motion trajectory are subjected to feature splicing to obtain a joint feature vector of both hands, wherein the joint feature vector of both hands is expressed as:

[0166] G=[v,log(PSD(f dominant ))]

[0167] Where G represents the joint feature vector of both hands, f dominant The main signal in the frequency domain signal representing the acceleration change sequence;

[0168] A binary classification judgment based on the action gesture or the motion trajectory is performed according to the joint feature vector of the two hands.

[0169] The palm change information represents the change difference between adjacent elements in the user's palm skeleton change sequence.

[0170] The energy spectral density is a physical quantity that describes the energy distribution of an energy signal in the frequency domain. It is defined as the square of the modulus of the Fourier transform of the signal and can be used to represent motion changes through energy changes.

[0171] The number of sampling points N represents the length of the discrete signal selected when performing Fourier transform. If the acceleration sensor collects 1000 data points per second, then a two-second acceleration change sequence has 2000 sampling points.

[0172] The feature concatenation is a process of weighted concatenation of the vectors of the gesture and motion trajectory. The two-handed joint feature vector is the concatenation result of the gesture and motion trajectory features and can be used as a multimodal feature.

[0173] The main signal refers to the frequency interval corresponding to the main frequency characteristic information.

[0174] The binary classification judgment refers to judging whether an action gesture occurs and whether a motion trajectory appears through a neural network.

[0175] Specifically, the present invention quantifies gestures and motion trajectories and performs feature concatenation to generate a combined feature vector for both hands. This allows for simultaneous monitoring of both left and right hand movements during detection. When the idle hand exhibits movement amplitude, the gesture v is given a greater weight, while also not neglecting subtle movements of the operating hand. This allows for simultaneous adjustment of the grinder's two functions when both the idle hand and the operating hand are in motion.

[0176] In the embodiment of the present invention, when the action gesture and motion trajectory are detected, it can be known that the user has performed a certain action instruction. However, the meaning of the action gesture and motion trajectory is still unknown, so subsequent steps are required for verification.

[0177] When the action gesture or the motion trajectory is not detected, the method returns to the above step of using the pre-built acceleration sensor to monitor the acceleration of the multifunctional grinder to obtain an acceleration change sequence.

[0178] Specifically, in the embodiment of the present invention, if no action gesture or the motion trajectory is detected, it indicates that the user is in a stable operation process and does not need to adjust the function of the grinder temporarily.

[0179] When the action gesture or the motion trajectory is detected, S8, a pre-built association database for querying gesture motion-service functions is used to perform a function query on the action gesture or motion trajectory to obtain a target function, and the multi-functional grinder is used to execute the target function.

[0180] The association database for querying gesture motion-service function refers to a database that stores association information between user gestures and specific grinder instructions.

[0181] The function query is performed by performing a keyword query process using a database tool in the association database for querying gesture motion-service functions. The target function represents the grinding machine instruction corresponding to the action gesture or the motion trajectory.

[0182] Specifically, in an embodiment of the present invention, before using a pre-built database for querying gesture motion-service function association, the method further includes:

[0183] When starting the pre-built action entry service, obtaining the user's repeatedly executed actions on the multi-functional grinder;

[0184] Fitting the repeatedly executed action to obtain an input action;

[0185] Obtain the action name configured by the user, obtain the function link configured by the user, and construct a key-value pair relationship between the function link and the input action;

[0186] According to the action name, the key-value pair relationship is entered into a pre-built association database for querying gesture motion-service function.

[0187] The action entry service is a data interface on the mobile phone side of the associated database for querying gesture movement-service functions.

[0188] The repeatedly executed action refers to the demonstration steps in which the user's free hand and the operating hand are within 10 cm of the grinder, and the same action is demonstrated three times.

[0189] The fitting refers to performing weighted average calculation on the three actions in the repeated action to convert them into one action. The input action is the fitting result of the repeated action.

[0190] The action name is a subjective name given by the user to the input action. The function link refers to the calling method of each function instruction of the grinder.

[0191] The key-value pair is a data storage method that makes the function link and the input action unique in the database. The key-value pair relationship is the result of the key-value pair construction of the function link and the input action.

[0192] Specifically, in an embodiment of the present invention, this method allows users to configure motion trajectories and gestures that suit them based on their work experience, such as using a pre-built mobile app to start a motion recording service. The user then holds the grinder in one hand. When a prompt message appears on the phone, the operator or free hand can perform a motion demonstration within a 10cm radius around the grinder three times. The database used to query the gesture motion-service function association automatically fits the repeated motions to obtain recorded data. Furthermore, a key-value pair is constructed between the user-defined function link and the recorded data to obtain a key-value pair relationship.

[0193] Then, in order to facilitate the user to modify the subsequent input actions, the present invention configures a user-configured action name for the key-value pair relationship.

[0194] In detail, in an embodiment of the present invention, after determining whether the action gesture or the motion trajectory is detected, the method further includes:

[0195] Obtain the successfully detected action gestures or motion trajectories within a preset time period to obtain historical execution actions;

[0196] Identifying a change pattern of action execution in the historical execution actions, and performing fuzzy processing based on action amplitude on the input action in the key-value pair relationship according to the change pattern of action execution to obtain an updated input action;

[0197] According to the update entry action, the key-value pair relationship in the association database for querying gesture motion-service function is updated. Wherein, the preset time period is configured as two weeks.

[0198] The historical execution actions are the action gestures and motion trajectories corresponding to the grinding machine control instructions successfully executed by the user within two weeks.

[0199] The action execution change rule represents the user action trend between the previous and next action gestures or motion trajectories that have been successfully detected.

[0200] The fuzzy processing refers to a method of gradually simplifying the pre-stored input actions according to the user's action trend to increase the accuracy of gesture recognition. The updated input action refers to the fuzzy input action.

[0201] Specifically, in an embodiment of the present invention, the initial input actions are all standard actions set by the user. However, in actual work, over time, users may perform the input actions more quickly and with less effort, resulting in substandard actions. To adapt to this change and improve user adaptability, each parameter in the input action is fuzzified through a weighted average calculation, thereby updating the key-value pair relationship in the association database used to query gesture movements and service functions.

[0202] Specifically, in an embodiment of the present invention, the function switching of the grinder during use can be controlled through the above process. Furthermore, in an embodiment of the present invention, action detection rules for horizontal placement detection and weightlessness and falling detection can be added to the association database for querying gesture motion-service functions, and the grinder can be configured to switch by detecting horizontal placement and falling. When the user picks up the grinder from a horizontal table, the grinder enters the working state. When the grinder is placed on a horizontal surface and remains stationary for a preset time, such as 3 seconds, the power of the grinder motor stops. When it is detected that the grinder itself is in a weightless state, the grinder motor is automatically stopped to protect the motor in the grinder.

[0203] Therefore, in the embodiment of the present invention, in the process of identifying the motion trajectory, the above-mentioned special opening and closing scenarios also need to be taken into consideration.

[0204] To address the issues described in the background art, the present invention first transforms a multifunctional grinder by replacing traditional buttons with gesture recognition. To improve the user's work continuity and efficiency, the present invention employs dual recognition of both the operator hand and the idle hand. The operator hand refers to the hand holding the multifunctional grinder, while the idle hand refers to the hand holding the customer's finger. Dual recognition involves functionally configuring the idle hand's gestures and the operator hand's motion trajectory. Both dual recognition operations utilize a combination of sensors and artificial intelligence, significantly improving detection accuracy and thus increasing the flexibility of the grinder's functional changes. Therefore, the present invention can improve the efficiency of grinding machine function switching.

[0205] like Figure 2 , which is a functional module diagram of a multifunctional grinding machine control system based on gesture recognition provided by one embodiment of the present invention.

[0206] The multifunctional sander control system 100 based on gesture recognition described in the present invention can be installed in an electronic device. Depending on the functionality implemented, the multifunctional sander control system 100 based on gesture recognition can include a sander acquisition module 101, an information acquisition module 102, a user gesture recognition module 103, a sander trajectory recognition module 104, and a command control module 105. A module, also referred to as a unit, is a series of computer program segments that can be executed by an electronic device processor and perform a fixed function. These modules are stored in the electronic device's memory.

[0207] The grinder acquisition module 101 is used to acquire a multifunctional grinder, wherein the multifunctional grinder includes an acceleration sensor and an environment sensing device;

[0208] The information acquisition module 102 is configured to monitor the acceleration of the multifunctional grinder using the acceleration sensor to obtain an acceleration change sequence;

[0209] The user gesture recognition module 103 is configured to determine whether the acceleration change sequence is within a preset working floating range, and when it is determined that the acceleration change sequence is within the working floating range, use the environmental sensing device to obtain ambient environment data of the multifunctional grinder, perform a palm object recognition operation on the ambient environment data to obtain a palm object set, perform a screening operation on the palm object set based on the user's idle hand to obtain a user's idle hand object, perform key node monitoring on the user's idle hand object to obtain a user's palm skeleton change sequence, and perform gesture recognition on the user's palm skeleton change sequence based on a pre-built gesture database to obtain an action gesture;

[0210] The grinding machine trajectory recognition module 104 is configured to perform motion trajectory recognition on the acceleration change sequence to obtain a motion trajectory when it is determined that the acceleration change sequence is not within the working floating range;

[0211] The instruction control module 105 is used to return to the above-mentioned step of using the pre-built acceleration sensor to monitor the acceleration of the multi-functional grinder and obtain an acceleration change sequence when the action gesture or the motion trajectory is not detected, and when the action gesture or the motion trajectory is detected, use the pre-built association database for querying gesture motion-service functions to perform a function query on the action gesture or motion trajectory to obtain a target function, and use the multi-functional grinder to execute the target function.

[0212] In detail, the modules in the multifunctional grinding machine control system 100 based on gesture recognition in the embodiment of the present invention are used in the same manner as above. Figure 1The same technical means are used as the multifunctional grinder control method based on gesture recognition described in , and can produce the same technical effects, so they will not be repeated here.

[0213] like Figure 3 , which is a structural diagram of an electronic device for implementing a multifunctional grinder control method based on gesture recognition provided by an embodiment of the present invention.

[0214] The electronic device 1 may include a processor 10 , a memory 11 and a bus 12 , and may further include a computer program stored in the memory 11 and executable on the processor 10 , such as a multifunctional sander control method program based on gesture recognition.

[0215] The memory 11 includes at least one type of readable storage medium, including a flash memory, a mobile hard disk, a multimedia card, a card-type memory (e.g., SD or DX memory), a magnetic memory, a disk, an optical disk, etc. In some embodiments, the memory 11 may be an internal storage unit of the electronic device 1, such as a mobile hard disk of the electronic device 1. In other embodiments, the memory 11 may also be an external storage device of the electronic device 1, such as a plug-in mobile hard disk, a smart memory card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device 1. Furthermore, the memory 11 also includes an internal storage unit of the electronic device 1 and an external storage device. The memory 11 can not only be used to store application software and various types of data installed on the electronic device 1, such as the code of the multi-functional grinder control method program based on gesture recognition, but can also be used to temporarily store data that has been output or is to be output.

[0216] In some embodiments, the processor 10 may be composed of an integrated circuit, such as a single packaged integrated circuit, or a plurality of packaged integrated circuits with the same or different functions, including one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and a combination of various control chips. The processor 10 is the control core (Control Unit) of the electronic device, connecting the various components of the entire electronic device using various interfaces and circuits. It executes or runs programs or modules stored in the memory 11 (such as a multi-function grinder control method program based on gesture recognition) and calls data stored in the memory 11 to perform various functions of the electronic device 1 and process data.

[0217] The bus 12 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus 12 may be divided into an address bus, a data bus, a control bus, etc. The bus 12 is configured to enable communication between the memory 11 and at least one processor 10, etc.

[0218] Figure 3 Only the electronic device with components is shown, and it can be understood by those skilled in the art that Figure 3 The structure shown does not constitute a limitation on the electronic device 1 , and may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0219] For example, although not shown, the electronic device 1 may further include a power source (such as a battery) for powering the various components. Preferably, the power source may be logically connected to the at least one processor 10 via a power management device, thereby implementing functions such as charging management, discharging management, and power consumption management through the power management device. The power source may further include any components such as one or more DC or AC power sources, a recharging device, a power failure detection circuit, a power converter or inverter, a power status indicator, etc. The electronic device 1 may further include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be described in detail here.

[0220] Furthermore, the electronic device 1 may also include a network interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between the electronic device 1 and other electronic devices.

[0221] Optionally, the electronic device 1 may further include a user interface, which may be a display or an input unit (such as a keyboard). Optionally, the user interface may also be a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touch device. The display may also be appropriately referred to as a display screen or a display unit, which is used to display information processed in the electronic device 1 and to display a visual user interface.

[0222] The multifunctional grinder control method program based on gesture recognition stored in the memory 11 of the electronic device 1 is a combination of multiple instructions. When running in the processor 10, it can achieve the following:

[0223] Obtain a multifunctional grinder, wherein the multifunctional grinder includes an acceleration sensor and an environmental sensing device;

[0224] Using the acceleration sensor, the acceleration of the multifunctional grinder is monitored to obtain an acceleration change sequence;

[0225] Determining whether the acceleration change sequence is within a preset working floating range;

[0226] When it is determined that the acceleration change sequence is within the working floating range, using the environmental sensing device to obtain ambient environment data of the multifunctional grinder, performing a palm object recognition operation on the ambient environment data to obtain a palm object set, and performing a screening operation on the palm object set based on the user's free hand to obtain a user's free hand object;

[0227] Performing key node monitoring on the user's idle hand object to obtain a sequence of changes in the user's palm skeleton, and performing gesture recognition on the sequence of changes in the user's palm skeleton based on a pre-built gesture database to obtain an action gesture;

[0228] When it is determined that the acceleration change sequence is not within the working floating range, performing motion trajectory recognition on the acceleration change sequence to obtain a motion trajectory;

[0229] determining whether the action gesture or the motion trajectory is detected;

[0230] When the action gesture or the motion trajectory is not detected, returning to the above step of using the pre-built acceleration sensor to monitor the acceleration of the multifunctional grinder to obtain an acceleration change sequence;

[0231] When the action gesture or the motion trajectory is detected, a pre-built association database for querying gesture motion-service functions is used to perform a function query on the action gesture or motion trajectory to obtain a target function, and the multi-functional grinder is used to execute the target function.

[0232] Specifically, the specific implementation method of the processor 10 for the above instructions can refer to Figures 1 to 3 The description of the relevant steps in the corresponding embodiments will not be repeated here.

[0233] Furthermore, if the modules / units integrated in the electronic device 1 are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile. For example, the computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a mobile hard drive, a magnetic disk, an optical disk, a computer memory, or a read-only memory (ROM).

[0234] The present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program. When the computer program is executed by a processor of an electronic device, the computer program can implement:

[0235] Obtain a multifunctional grinder, wherein the multifunctional grinder includes an acceleration sensor and an environmental sensing device;

[0236] Using the acceleration sensor, the acceleration of the multifunctional grinder is monitored to obtain an acceleration change sequence;

[0237] Determining whether the acceleration change sequence is within a preset working floating range;

[0238] When it is determined that the acceleration change sequence is within the working floating range, using the environmental sensing device to obtain ambient environment data of the multifunctional grinder, performing a palm object recognition operation on the ambient environment data to obtain a palm object set, and performing a screening operation on the palm object set based on the user's free hand to obtain a user's free hand object;

[0239] Performing key node monitoring on the user's idle hand object to obtain a sequence of changes in the user's palm skeleton, and performing gesture recognition on the sequence of changes in the user's palm skeleton based on a pre-built gesture database to obtain an action gesture;

[0240] When it is determined that the acceleration change sequence is not within the working floating range, performing motion trajectory recognition on the acceleration change sequence to obtain a motion trajectory;

[0241] determining whether the action gesture or the motion trajectory is detected;

[0242] When the action gesture or the motion trajectory is not detected, returning to the above step of using the pre-built acceleration sensor to monitor the acceleration of the multifunctional grinder to obtain an acceleration change sequence;

[0243] When the action gesture or the motion trajectory is detected, a pre-built association database for querying gesture motion-service functions is used to perform a function query on the action gesture or motion trajectory to obtain a target function, and the multi-functional grinder is used to execute the target function.

[0244] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, systems and methods can be implemented in other ways. For example, the system embodiments described above are only exemplary, and actual implementations may have other division methods.

[0245] The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed across multiple network elements. Some or all of the modules may be selected to achieve the purpose of the solution of this embodiment according to actual needs.

[0246] In addition, the functional modules in various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or hardware plus software functional modules.

[0247] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0248] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A multifunctional grinding machine control method based on gesture recognition, characterized in that: The method comprises: Obtain a multifunctional grinder, wherein the multifunctional grinder includes an acceleration sensor and an environmental sensing device; Using the acceleration sensor, the acceleration of the multifunctional grinder is monitored to obtain an acceleration change sequence; Determining whether the acceleration change sequence is within a preset working floating range; When it is determined that the acceleration change sequence is within the working floating range, using the environmental sensing device to obtain ambient environment data of the multifunctional grinder, performing a palm object recognition operation on the ambient environment data to obtain a palm object set, and performing a screening operation on the palm object set based on the user's free hand to obtain a user's free hand object; Performing key node monitoring on the user's idle hand object to obtain a sequence of changes in the user's palm skeleton, and performing gesture recognition on the sequence of changes in the user's palm skeleton based on a pre-built gesture database to obtain an action gesture; When it is determined that the acceleration change sequence is not within the working floating range, performing motion trajectory recognition on the acceleration change sequence to obtain a motion trajectory; determining whether the action gesture or the motion trajectory is detected; When the action gesture or the motion trajectory is not detected, returning to the above step of using the pre-built acceleration sensor to monitor the acceleration of the multifunctional grinder to obtain an acceleration change sequence; When the action gesture or the motion trajectory is detected, a pre-built association database for querying gesture motion-service functions is used to perform a function query on the action gesture or motion trajectory to obtain a target function, and the multi-functional grinder is used to execute the target function.

2. The multifunctional grinding machine control method based on gesture recognition according to claim 1, characterized in that: The performing a palm object recognition operation on the surrounding environment data to obtain a palm object set includes: Performing Gaussian filtering on the surrounding environment data to obtain noise-reduced environmental data; Performing an edge recognition operation on the noise reduction environment data to obtain an object set; A preset palm type classification and judgment operation is performed on each object in the object set to obtain a palm set.

3. The multifunctional grinding machine control method based on gesture recognition according to claim 2, characterized in that: The filtering operation based on the user's free hand on the palm object set to obtain the user's free hand object includes: performing palm gesture recognition on each palm object in the palm object set to obtain a palm gesture feature set; According to the palm gesture feature set, a type recognition operation is performed on the palm object set to obtain a grinder execution hand and a customer target hand; The palm objects other than the grinder execution hand and the customer target hand in the palm object set are determined as user idle hand objects.

4. The multifunctional grinding machine control method based on gesture recognition according to claim 3, characterized in that: The determining whether the action gesture or the motion trajectory is detected includes: Get an action gesture, where the action gesture is represented by: v=F[Δx1,Δy1,Δz1,…,Δx k ,Δy k ,Δz k ] Where v represents the action gesture, F[·] represents gesture recognition, Δx1, Δy1, Δz1 represent the palm information changes of the first element in the user's palm skeleton change sequence in the three dimensions of x, y, and z, Δx k ,Δy k ,Δz k Indicates the palm information change of the kth element in the user's palm skeleton change sequence in the x, y, and z dimensions; Obtain a motion trajectory, wherein the motion trajectory is expressed as: Where PSD(f) represents the motion trajectory, is the energy spectrum density in the frequency domain signal of the acceleration change sequence, f represents the frequency, a(t) represents the acceleration change sequence, e -j2πft / N represents the complex exponential function, j represents the imaginary unit, t represents the time, and N represents the number of sampling points; The action gesture and motion trajectory are subjected to feature splicing to obtain a joint feature vector of both hands, wherein the joint feature vector of both hands is expressed as: G=[v,log(PSD(f dominant ))] Where G represents the joint feature vector of both hands, f dominant The main signal in the frequency domain signal representing the acceleration change sequence; A binary classification judgment based on the action gesture or the motion trajectory is performed according to the joint feature vector of the two hands.

5. The multifunctional grinding machine control method based on gesture recognition according to claim 4, characterized in that: The step of monitoring key nodes of the user's idle hand object to obtain a user palm skeleton change sequence includes: Performing a feature extraction operation on the user's idle hand object to obtain a palm feature distribution set, and performing a feature weight sorting operation based on principal component feature analysis on the palm feature distribution set to obtain a palm key feature set; In the user's idle hand object, performing position coordinate selection on each palm key feature in the palm key feature set to obtain a palm key node set; Performing palm skeleton connection on the palm key node set to obtain a palm point-line skeleton graph; The change path of the palm point-line skeleton diagram within a preset time period is recorded to obtain a user palm skeleton change sequence.

6. The multifunctional grinding machine control method based on gesture recognition according to claim 5, characterized in that: The step of performing gesture recognition on the user's palm skeleton change sequence based on a pre-built gesture database to obtain an action gesture includes: According to a preset amplitude threshold, a filtering operation based on the movement amplitude is performed on the user's palm skeleton change sequence to obtain a set of fingers with the largest amplitude movement; acquiring, from the user's palm skeleton change sequence, finger movements of each finger with the largest amplitude movement in the set of fingers with the largest amplitude movement, to obtain a finger skeleton change set; performing recognition operations based on single finger movements and coordinated finger movements on the finger skeleton change set to obtain a finger movement set; According to the pre-built gesture database, a similarity clustering and screening operation is performed on the finger action set to obtain action gestures.

7. The multifunctional grinding machine control method based on gesture recognition according to claim 6, characterized in that: The step of performing motion trajectory recognition on the acceleration change sequence to obtain the motion trajectory includes: Extracting waveform features of the acceleration change sequence to obtain waveform amplitude information and waveform duration; Performing a Fourier transform operation on the acceleration change sequence to obtain frequency domain signal features, and performing frequency feature extraction on the frequency domain signal features to obtain main frequency feature information and period feature information; Key motion trajectory identification is performed on the waveform amplitude information, waveform duration, main frequency characteristic information and period characteristic information to obtain a motion trajectory.

8. The multifunctional grinding machine control method based on gesture recognition according to claim 7, characterized in that: Before using the pre-built database for querying gesture motion-service function association, the method further includes: When starting the pre-built action entry service, obtaining the user's repeatedly executed actions on the multi-functional grinder; Fitting the repeatedly executed action to obtain an input action; Obtain the action name configured by the user, obtain the function link configured by the user, and construct a key-value pair relationship between the function link and the input action; According to the action name, the key-value pair relationship is entered into a pre-built association database for querying gesture motion-service function.

9. The multifunctional grinding machine control method based on gesture recognition according to claim 8, characterized in that: After determining whether the action gesture or the motion trajectory is detected, the method further includes: Obtain the successfully detected action gestures or motion trajectories within a preset time period to obtain historical execution actions; Identifying a change pattern of action execution in the historical execution actions, and performing fuzzy processing based on action amplitude on the input action in the key-value pair relationship according to the change pattern of action execution to obtain an updated input action; According to the update entry action, the key-value pair relationship in the association database for querying gesture motion-service function is updated.

10. A multifunctional grinding machine control system based on gesture recognition, characterized in that: The system comprises: A grinder acquisition module, configured to acquire a multifunctional grinder, wherein the multifunctional grinder includes an acceleration sensor and an environmental sensing device; An information acquisition module, configured to monitor the acceleration of the multifunctional grinder using the acceleration sensor to obtain an acceleration change sequence; a user gesture recognition module, configured to determine whether the acceleration change sequence is within a preset working floating range, and when it is determined that the acceleration change sequence is within the working floating range, using the environmental sensing device to obtain ambient environment data of the multifunctional grinder, performing a palm object recognition operation on the ambient environment data to obtain a palm object set, performing a screening operation on the palm object set based on the user's idle hand to obtain a user's idle hand object, and performing key node monitoring on the user's idle hand object to obtain a user's palm skeleton change sequence, and performing gesture recognition on the user's palm skeleton change sequence based on a pre-built gesture database to obtain an action gesture; a grinding machine trajectory recognition module, configured to, when determining that the acceleration change sequence is not within the working floating range, perform motion trajectory recognition on the acceleration change sequence to obtain a motion trajectory; The instruction control module is used to, when the action gesture or the motion trajectory is not detected, return to the above-mentioned step of using the pre-built acceleration sensor to monitor the acceleration of the multi-functional grinder to obtain an acceleration change sequence; and when the action gesture or the motion trajectory is detected, use a pre-built association database for querying gesture motion-service functions to perform a function query on the action gesture or motion trajectory to obtain a target function, and use the multi-functional grinder to execute the target function.

Images