Fingertip Detection Using Depth Minimum and Hand Apex
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing fingertip detection methods are limited by gesture posture and position, leading to inaccurate and inefficient detection, especially when the finger is facing a camera, and are affected by noise interference and slow calculation speeds.
Innovation Solution
A method that uses depth image processing to determine a minimum-depth point and hand vertex, extracting an effective foreground depth map to accurately detect the fingertip position regardless of gesture orientation, improving robustness and efficiency by filtering out noise and background interference.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the fingertip algorithm based on geodesic distances is used, then the detection can be performed, but the detection is limited by postures and positions of gesture and cannot detect fingertip when finger is facing camera
Solution Approach 1:
The patent transforms the 2D image coordinate problem into a 3D depth space problem by introducing depth information. Instead of relying on 2D geometric features that fail when fingers face the camera, the method uses depth values to identify fingertips as minimum-depth points along hand contour lines, enabling detection in three-dimensional space regardless of gesture orientation.
Solution Approach 2:
The patent changes the detection parameter from 2D geometric coordinates to 3D depth values. By using depth information as the primary parameter for fingertip detection and introducing the concept of hand contour lines in depth space, the method achieves robust detection across various gesture postures and positions that were previously undetectable.
2Measurement precision
If the coordinate algorithm based on geometric features is used, then the detection can be performed, but the calculation speed is slow and calculation effect is poor
Solution Approach 1:
The patent extracts only the essential depth information along hand contour lines rather than processing entire hand regions or performing complex geometric calculations. By focusing extraction on minimum-depth points along contours, the method achieves high precision fingertip positioning with significantly reduced computational burden.
Solution Approach 2:
The patent replaces complex geometric calculation mechanisms with simpler depth-based comparison mechanisms. Instead of calculating geometric features and their relationships in 2D space, the method uses direct depth value comparisons in 3D space, substituting computationally intensive operations with simpler arithmetic operations that maintain precision while improving speed.
3Measurement precision
If traditional fingertip detection methods are used, then detection can be performed, but noise interference affects detection accuracy
Solution Approach 1:
The patent introduces hand contour lines as an intermediary structure that separates the fingertip detection task from noisy background information. By constraining the search for fingertips to along these contour lines in depth space, the method effectively filters out noise and background interference while maintaining detection accuracy.
Data Source
Figure 1A~1C
Figure 2
Figure 3
AI summary
Disclosed are a fingertip detection method, a fingertip detection means, a fingertip detection device, and a medium. The fingertip detection method comprises: determining a minimum-depth point (N) in a hand region to be detected on the basis of a depth map; determining a hand apex (T) in said hand region on the basis of the depth map; and determining the position of a fingertip on the basis of the minimum-depth point (N) and the hand apex (T). On the basis of the minimum-depth point (N) and the hand apex (T), in the case where a user makes different gestures, the fingertip can be accurately and efficiently detected.