A gesture recognition method and device
A gesture recognition and gesture technology, applied in character and pattern recognition, instruments, computing, etc., can solve the problems of reduced recognition rate and no support for hand rotation without obvious motion trajectories.
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0047] This embodiment introduces an exemplary system for running the gesture recognition method proposed by the present invention, such as figure 2 As a schematic diagram of the system structure, the system implements an example of a third-party application program using the present invention for human-computer interaction.
[0048] The system acquires the acceleration data sequence and angular velocity data sequence of the user's gesture operation through a device with built-in acceleration and angular velocity sensors (ie data recognition unit 210), and preprocesses the acceleration data sequence and angular velocity data sequence via Bluetooth or its transmission to other connected devices Module 220;
[0049] The preprocessing module 220 performs denoising and smoothing processing on the acquired acceleration data sequence and angular velocity data sequence, normalizing the denoising and smoothing result, and then performing the dimension standardization operation on the norma...
Embodiment 2
[0053] This embodiment introduces the process of preprocessing the original acceleration data sequence and angular velocity data sequence by the preprocessing module 220. In this embodiment, the original data sequence of a user gesture includes six, denoted as AccSeq_ x ,AccSeq_ y ,AccSeq_ z ,AngSeq_ x ,AngSeq_ y ,AngSeq_ z ; Among them, AccSeq represents the acceleration data sequence, AngSeq represents the angular velocity data sequence, and the subscripts x, y, and z respectively represent the three direction axes of the sensor.
[0054] Such as image 3 The implementation flowchart of this embodiment includes the following steps:
[0055] Step 301: Denoising and smoothing. This embodiment adopts mean filtering, and takes a sliding window with a width of 5 as the center point area, and uses the mean value of the data in the window as the updated value of the center point to perform mean filtering. The calculation formula is:
[0056] among them,
[0057] Value i+j Represents the...
Embodiment 3
[0081] This embodiment introduces the process of using the recognition module 230 to perform gesture recognition on the feature vector obtained after preprocessing. The identification module 230 includes a predefined classifier 231 and a custom classifier 232 set by the user. Firstly, the predefined classifier 231 performs gesture recognition. When the predefined classifier 231 recognizes the gesture input by the user as one of the 38 predefined gestures, exit the module and return to the third-party application or the result of the user recognition; When the predefined classifier 231 cannot be identified or is recognized as a negative sample, the custom classifier 232 will continue to identify whether the current input gesture is a user-defined gesture. For detailed operations, see Figure 5 ,Proceed as follows.
[0082] Step 501: The predefined classifier 231 recognizes the feature vector of the user input gesture. In consideration of user experience, this embodiment may use a...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More 


