3D Gesture Recognition Module Using Area Variation Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing gesture recognition systems using 3D images face challenges in accurately recognizing hand gestures due to the fuzziness of details, which reduces the accuracy of gesture recognition compared to 2D images.
Innovation Solution
A gesture recognition module and method that calculates area variations of the user's hand in 3D images by capturing coordinates, defining zones, and calculating areas to recognize gestures, thereby improving the accuracy of gesture recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If 3D images are used for gesture recognition, then depth information can be utilized to exclude unnecessary image information and increase recognition accuracy, but the fuzziness of details in 3D images makes it difficult to use them as references for recognizing gestures
Solution Approach 1:
The patent divides the hand into multiple zones (first zone, second zone, etc.) based on coordinate information, and calculates area variations for each zone separately. This segmentation allows the system to handle the fuzziness of 3D images by focusing on specific regions and their area changes rather than relying on overall detail clarity.
Solution Approach 2:
The patent transitions from using only coordinate information to incorporating area information as an additional dimension. By calculating the area of zones defined by coordinates and using area variations alongside coordinate changes for gesture recognition, the system compensates for the detail fuzziness in 3D images through this new dimensional feature.
2Difficulty of detecting and measuring
If 2D images are used for gesture recognition, then detailed information such as finger details can be clearly shown, but depth information is lost which reduces the ability to exclude unnecessary image information
Solution Approach 1:
The patent merges the advantages of both 2D and 3D images by combining coordinate information (from 3D depth data) with area calculations (derived from 2D image projections). This fusion allows the system to benefit from both the depth information for excluding unnecessary data and the area variations for maintaining detail awareness.
Solution Approach 2:
The patent uses area information as an intermediary feature that bridges the gap between 2D detail clarity and 3D depth information. By calculating areas of zones defined by 3D coordinates and using these area variations as a reference, the system can recognize gestures accurately without requiring either pure 2D or pure 3D images alone.
Data Source
AI summary
A gesture recognition module, for recognizing a gesture of a user, includes a detecting unit, for capturing at least one hand image of a hand of the user, so as to sequentially acquire a first coordinate and a second coordinate; a computing unit, coupled to the detecting unit for defining a first zone and a second zone according to the first coordinate and the second coordinate, respectively, and calculating a first area and a second area according to the first zone and the second zone; and a determining unit, coupled to the detecting unit and the computing unit for recognizing the gesture according to the first coordinate, the second coordinate, the first area and the second area.