3D Depth Camera Gesture Recognition via Center Point Velocity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing hand gesture recognition techniques using image cameras are complex and struggle to stably detect hands in various shapes and circumstances, requiring extensive algorithms and candidate image comparisons.
Innovation Solution
A device and method utilizing a 3D depth camera to acquire and process video, calculating the 3D center point of a gesture object, and recognizing gestures based on velocity differences between motion coordinates, simplifying the algorithm by binarizing frames and detecting blobs to determine center points and velocities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If traditional image camera and linear tracking method are used, then gesture recognition can be implemented, but the algorithm becomes complex and difficult to implement
Solution Approach 1:
The patent changes the detection parameter from 2D image coordinates to 3D depth coordinates. By using a depth camera to acquire Z-axis information and calculating 3D center points, the system achieves more reliable gesture detection with simpler algorithms, as the depth information provides direct spatial separation between hand and background
Solution Approach 2:
The patent adds the depth dimension (Z-axis) to the traditional 2D image processing. By acquiring 3D coordinates including depth information and calculating 3D center points, the system transforms the problem from 2D trajectory tracking to 3D spatial analysis, which simplifies the algorithm while improving reliability
2Productivity
If candidate region comparison method is used, then hand detection can be performed, but it requires acquiring and comparing all candidate images which increases computational complexity
Solution Approach 1:
The patent extracts only the essential feature (3D center point coordinates) from the gesture object rather than processing entire candidate regions. By calculating the 3D center point from depth camera data and tracking its motion, the system achieves fast processing without the need to acquire and compare multiple candidate images
Solution Approach 2:
The patent replaces the mechanical process of acquiring and comparing multiple candidate images with a direct calculation approach. Instead of iterating through candidate regions, the system directly computes 3D center points from depth data and compares coordinate differences, significantly reducing computational complexity
3Measurement precision
If 3D depth camera and center point velocity method are used, then gesture recognition accuracy increases, but the system requires 3D depth camera hardware
Solution Approach 1:
The patent introduces the 3D center point as an intermediary representation. Instead of directly analyzing complex hand shapes or comparing candidate images, the system uses the 3D center point coordinates as a simplified mediator that captures the essential motion information, achieving high accuracy while requiring only depth camera hardware
Data Source
AI summary
Provided are a device and method for recognizing a gesture. The device includes a video image acquirer configured to acquire a video including a gesture object by photographing the gesture object, a center point detector configured to detect a center point of the gesture object in frames of the acquired video, a velocity calculator configured to calculate a velocity of the gesture object by calculating a difference between motion coordinates on each axis on the basis of the center point of the gesture object, and a gesture recognizer configured to recognize a gesture of the gesture object by comparing the calculated velocity of the gesture object with a preset velocity.