Head-Mounted Gesture Recognition with Region Motion Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing gesture recognition technologies in electronic devices like smart glasses struggle with accuracy in various use situations, particularly in low-light conditions or when users wear gloves, leading to unstable gesture detection.
Innovation Solution
A gesture recognition apparatus utilizing an imaging unit, system control unit, and gesture recognition unit that includes a target-portion detection, important-region detection, motion detection, and gesture detection units to accurately identify hand positions and joint movements, employing image processing and deep learning for precise gesture recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If gesture recognition is performed using conventional image processing methods, then the system can operate in various situations, but the recognition accuracy becomes unstable in low-light conditions or when users wear gloves
Solution Approach 1:
The patent transitions from 2D image processing to 3D depth information processing by introducing a depth acquisition unit. This dimensional change enables the system to capture spatial relationships and hand shapes in three dimensions, allowing accurate gesture recognition even when visual features are obscured by gloves or low light conditions, as depth information provides structural data independent of surface appearance
Solution Approach 2:
The patent changes the fundamental parameter used for gesture recognition from 2D visual features (color, texture, brightness) to 3D geometric parameters (depth, spatial coordinates, hand shape). This parameter transformation allows the system to maintain recognition accuracy across varying lighting conditions and glove types, as depth parameters remain consistent regardless of surface appearance changes
2Measurement precision
If deep learning is used to recognize gestures, then recognition capability is enhanced, but the system still fails to maintain high accuracy in challenging environments like dark locations or with gloved hands
Solution Approach 1:
The patent introduces depth information as an intermediary that bridges the gap between the imaging unit and the gesture recognition process. This depth data serves as a mediator that provides robust spatial and structural information about hand gestures, enabling the recognition system to maintain high accuracy and reliability even when traditional visual features are unavailable or obscured by environmental factors
Solution Approach 2:
The patent segments the gesture recognition process into distinct functional units: a depth acquisition unit that captures 3D information, and a gesture recognition unit that processes this depth data. This segmentation allows the system to specialize each unit for its optimal function, with the depth acquisition unit providing reliable spatial data independent of lighting conditions, thereby enhancing overall system reliability
3Ease of operation
If joint detection and hand shape specification are performed in 2D images, then the system can identify gestures, but detection accuracy deteriorates when key points are obscured or visibility is poor
Solution Approach 1:
The patent moves joint detection from 2D image space to 3D depth space by introducing a depth acquisition unit. This dimensional transition enables the system to detect hand joints and key points based on their spatial coordinates and depth information, maintaining high detection accuracy even when 2D visual features are obscured by gloves, shadows, or poor lighting conditions
Data Source
Figure 1
Figure 2
Figure 3A~3B
AI summary
A gesture recognition apparatus according to the present invention includes: a portion detection means configured to detect from a captured image a portion making a gesture; a recognition means configured to recognize the gesture on a basis of motion of the portion detected by the portion detection means; a region detection means configured to detect from the portion a region to be used for the gesture; and a motion detection means configured to detect motion of the region detected by the region detection means, wherein the recognition means recognizes the gesture on a basis of the motion detected by the motion detection means.