Fingernail Tracking via Segmentation and Shape Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing feature tracking systems lack the accuracy and processing efficiency required for real-time tracking and augmentation of fingernails in captured image data, which is necessary for robust and efficient finger nail tracking.
Innovation Solution
A method and system that utilize feature detection functions like pixel colour segmentation, Haar wavelets, shape models from machine learning, and depth analysis to estimate and refine the location of fingernails, incorporating a tracking phase with translation displacement calculation and refinement using digit-shape and sub-shape models, along with cascading regression coefficient matrices for improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If pixel colour segmentation is used for finger detection, then processing speed is improved, but measurement precision of fingernail location deteriorates
Solution Approach 1:
The patent segments the finger detection process into two distinct phases: a rough localization stage using pixel color segmentation to identify candidate regions, and a refinement stage using shape models and regression analysis to precisely locate fingernails. This segmentation allows each method to operate in its optimal performance range.
Solution Approach 2:
The patent performs preliminary rough localization of fingernails using computationally efficient pixel color segmentation before applying more computationally intensive refinement methods. This preliminary action narrows down the search space and enables subsequent precise localization to focus only on relevant regions.
2Productivity
If simple pixel colour segmentation is used, then processing efficiency is improved, but tracking accuracy deteriorates
Solution Approach 1:
The tracking process is segmented into initialization using pixel color segmentation and subsequent refinement using shape models with regression coefficient matrices. This allows the system to maintain high processing efficiency while achieving robust tracking accuracy through the refinement stage.
Solution Approach 2:
The patent implements feedback mechanisms where detected fingernail positions are continuously refined using shape models and regression analysis. The system uses feedback from the rough detection to guide the refinement process, improving tracking accuracy while maintaining efficiency through iterative optimization.
3Adaptability or versatility
If feature tracking systems are used, then object tracking capability is improved, but measurement precision of fingernail features deteriorates
Solution Approach 1:
The patent applies local quality by using specialized shape models and regression coefficient matrices specifically designed for fingernail geometry. Instead of applying generic tracking algorithms uniformly, the system uses tailored models that capture the specific structural characteristics of fingernails, thereby improving measurement precision for these features.
Solution Approach 2:
The patent transforms the tracking problem by changing parameters from generic object tracking to specific fingernail feature tracking using shape models. The regression coefficient matrices adjust key parameters like position, orientation, and scale specifically for fingernail geometry, improving measurement precision while maintaining tracking capability.
Data Source
AI summary
A computer-implemented method and system are described for locating fingernails of a person's hand in an image. An approximate location of each fingernail in the image is determined. An approximate location of each of a plurality of digit-shape objects in the image is initialised based on the approximate locations of the fingernails, and initially refined based on respective digit-shape object models and corresponding functions. The approximate location of each fingernail sub-shape is further refined based on a respective fingernail model and its corresponding function.