Real-Time Hand Tracking via Distance Transform Palm Center Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing hand gesture recognition systems are cumbersome, costly, and sensitive to luminosity variations and background changes, making them inefficient for real-time hand movement tracking without the need for cumbersome gloves.
Innovation Solution
A method for real-time hand movement tracking that locates hand contours and extracts postural characteristics using a distance transform to find the center of the palm, determines fingertip and finger base positions, and identifies the thumb based on angle and length criteria, without requiring assumptions about hand configuration or wearing gloves.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If electronic gloves with sensors are used to determine hand position and finger joint angles, then hand gesture recognition accuracy is improved, but device complexity and cost increase
Solution Approach 1:
The patent creates a virtual copy of the hand by detecting hand contours and extracting postural characteristics from video images. Instead of using physical sensors on the hand, the system captures optical information from a video camera, processes it through contour detection and distance transform algorithms, and generates a digital representation of hand position and posture that can be used for gesture recognition, thereby eliminating the need for complex electronic gloves
Solution Approach 2:
The patent replaces the mechanical sensor-based system (electronic gloves with physical sensors) with an optical-computational system. The solution uses video camera imaging combined with image processing algorithms (contour detection, distance transform, postural characteristic extraction) to achieve hand tracking, substituting physical mechanical sensors with optical fields and computational methods
2Measurement precision
If colored gloves are used to locate hand contours in images, then hand detection is improved, but sensitivity to luminosity variations and background changes increases
Solution Approach 1:
The patent enables the hand to be detected without requiring external assistance such as colored gloves. The system uses the hand's natural appearance in video images, detecting contours through image processing algorithms that analyze pixel intensity variations and spatial relationships. The distance transform and postural characteristic extraction methods allow the system to identify hand contours based on their geometric properties rather than color, making the detection self-sufficient and independent of artificial markers
Solution Approach 2:
The patent changes the detection parameter from color-based identification (colored gloves) to geometric and spatial parameter-based identification. By using distance transform to calculate distances from each pixel to the hand contour and identifying postural characteristics through geometric relationships, the system achieves luminosity-invariant hand detection that is not affected by lighting conditions or background color variations
3Measurement precision
If center of gravity calculation is used to determine hand center position, then hand positioning is improved, but accuracy decreases when forearm is included in the region
Solution Approach 1:
The patent extracts only the relevant portion of the hand region for center position calculation. Instead of using the entire region including forearm, the system first identifies hand contours, then performs distance transform specifically on the hand region to find the pixel furthest from the contour. This extraction approach isolates the palm center calculation from the forearm, ensuring accurate hand center positioning even when the forearm is present in the image
Solution Approach 2:
The patent segments the hand detection process into distinct stages: contour detection, distance transform application to the hand region, and postural characteristic extraction. By segmenting the region of interest and applying specific algorithms to each segment, the system accurately distinguishes the hand center from the forearm, maintaining positioning accuracy regardless of forearm inclusion in the overall image region
4Measurement precision
If pattern initialization with hand configuration hypotheses is used to follow hand movements, then movement tracking is improved, but system complexity and initialization difficulty increase
Solution Approach 1:
The patent performs preliminary extraction of postural characteristics from each video frame, including contour detection, distance transform calculation, and identification of key hand points (palm center, fingertips, finger bases). This preliminary processing establishes the hand's geometric configuration without requiring hypotheses about hand posture, enabling straightforward movement tracking by comparing extracted characteristics between consecutive frames
Solution Approach 2:
Instead of initializing a pattern model with hypotheses about hand configuration and then trying to fit it to the image, the patent inverts the approach by directly extracting hand postural characteristics from the image itself through contour detection and distance transform. This inversion eliminates the need for pattern initialization hypotheses, as the system derives hand configuration directly from observed image data rather than from pre-defined models
Data Source
AI summary
A method for following hand movements in an image flow, includes receiving an image flow in real time, locating in each image in the received image flow a hand contour delimiting an image zone of the hand, extracting the postural characteristics from the image zone of the hand located in each image, and determining the hand movements in the image flow from the postural characteristics extracted from each image. The extraction of the postural characteristics of the hand in each image includes locating in the image zone of the hand the center of the palm of the hand by searching for a pixel of the image zone of the hand the furthest from the hand contour.