Motion Recognition Using Orthogonal Semi-Supervised NMF
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing motion recognition systems for mobile devices face challenges in accurately interpreting sensor data due to variations in user motion and device positioning, leading to inconsistent results.
Innovation Solution
An apparatus and method utilizing Orthogonal Non-negative Matrix Factorization (ONMF) and Orthogonal Semi-Supervised Non-negative Matrix Factorization (OSSNMF) to extract feature vectors from sensor data, integrating frequency and time domain data, and calibrating sensor data to classify user motions effectively, employing Stiefel manifolds for orthogonalization and reducing resource requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional sensor data interpretation methods are used, then the system is simple to implement, but the motion recognition accuracy deteriorates due to variations in user motion and device positioning
Solution Approach 1:
The patent transforms sensor data from time domain to frequency domain using Fast Fourier Transform (FFT), changing the representation parameters of the data. This transformation enables the extraction of motion features that are invariant to device positioning and user motion variations, thereby improving motion recognition accuracy without proportionally increasing system complexity
Solution Approach 2:
The patent replaces conventional motion recognition algorithms with Orthogonal Non-negative Matrix Factorization (ONMF) and Orthogonal Semi-Supervised Non-negative Matrix Factorization (OSSNMF) methods. These mathematical decomposition techniques substitute traditional approaches, providing more robust feature extraction that handles variations in user motion and device positioning while maintaining computational feasibility
2Measurement precision
If comprehensive sensor data processing is applied to improve recognition accuracy, then motion classification improves, but computational resource consumption increases
Solution Approach 1:
The patent extracts only the most relevant motion features from sensor data by applying ONMF and OSSNMF decomposition. This selective extraction process identifies and isolates key frequency domain characteristics that are discriminative for motion classification, reducing the amount of data that needs to be processed while maintaining high classification accuracy and lowering computational resource consumption
3Reliability
If frequency domain analysis is used to improve motion feature extraction, then recognition robustness improves, but processing time increases
Solution Approach 1:
The patent performs Fast Fourier Transform (FFT) to convert sensor data to frequency domain as a preliminary processing step before feature extraction and classification. This pre-processing transformation is computationally efficient and enables subsequent ONMF/OSSNMF decomposition to proceed more quickly, achieving robust motion feature extraction while minimizing additional processing time
Data Source
AI summary
An apparatus for recognizing a user's motions based on sensor information and label information, a method for establishing an ONMF-based basis matrix, and a method for establishing an OSSNMF-based basis matrix are provided, where the basis matrices are used to extract motion features of the user. The apparatus for recognizing the user's motions may include a feature vector extractor configured to multiply a transposed matrix of an orthogonalized basis matrix by a sensor data matrix of frequency domain sensor data acquired from sensors to extract an ONMF-based feature vector and a multi-class classifier configured to use the extracted ONMF-based feature vector to classify the user's motion into a type from among types of a user's motions.