Dynamic Gesture Recognition Using Event-Driven Vision Sensors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional human-machine interaction systems based on static gesture recognition suffer from long response times, high computation burden, large power consumption, and complex installation, limiting their effectiveness and usability.
Innovation Solution
A dynamic gesture identification method utilizing a Dynamic Vision Sensor (DVS) to process event data flow, enabling rapid hand detection and tracking through event cumulative graphs, without image processing, and employing Kalman filtering and neural networks for gesture recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If conventional static gesture recognition methods are used, then gesture identification can be achieved, but response time is long and computation burden is high
Solution Approach 1:
The patent replaces conventional frame-based image processing with event-driven processing from Dynamic Vision Sensors. Instead of processing complete images at fixed frame rates, the system processes only changed pixel events asynchronously, eliminating frame rate limitations and reducing computation burden while improving response time
Solution Approach 2:
The patent extracts only the essential hand position information from event data using simple statistical calculations (mean and variance of event coordinates) rather than processing complete images. This extraction approach achieves adequate gesture recognition with minimal computation, dramatically reducing processing time and computational requirements
2Measurement precision
If conventional image processing methods are used for gesture recognition, then accurate identification can be achieved, but power consumption is high
Solution Approach 1:
The patent substitutes energy-intensive frame-based image processing with efficient event-driven processing. The DVS architecture generates events only when changes occur, and the processing involves simple coordinate statistics rather than full image analysis, achieving comparable accuracy with significantly reduced power consumption
Solution Approach 2:
The patent uses only the necessary portion of visual information (hand position events) rather than processing complete images. By focusing computation only on relevant changed regions and using minimal processing per event, the system achieves sufficient accuracy while consuming far less power than conventional methods
3Adaptability or versatility
If conventional human-machine interaction systems are used, then interaction functions can be achieved, but installation is complicated
Solution Approach 1:
The patent uses a DVS that can directly capture dynamic visual events without requiring integration with specific hardware platforms. The event data format is universal and can be processed by any system implementing the described algorithms, simplifying installation and improving adaptability across different applications and devices
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
The present disclosure provides a dynamic gesture identification method, a gesture interaction method and an interaction system. The interaction system includes: a dynamic vision sensor configured to trigger an event in accordance with movement of an object in a field of view relative to the dynamic vision sensor, and output an event data flow; a hand detection module configured to process the received event data flow to determine an initial position of a hand; a hand tracking module configured to determine a series of state vectors in accordance with the initial position of the hand; a gesture identification module configured to create an event cloud in accordance with event data corresponding to the obtained state vector, and process the event cloud to identify the gesture; and a command execution module configured to execute a corresponding operating command in accordance with the identified gesture.