Indoor Activity Recognition via PCA Signal Denoising
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing indoor non-contact human activity recognition systems face challenges in achieving high-precision recognition with minimal data samples, often resulting in over-fitting or imprecise results due to redundant data and complex processing times.
Innovation Solution
The method involves collecting indoor reflected signals using an antenna array, filtering them through principal component analysis to remove noise, and inputting the cleaned signals into a pre-trained CNN model based on a transfer learning algorithm, which is trained using a simplified structure and optimized for efficient data usage across different positions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a model is trained with redundant data, then the training process can be completed, but over-fitting is caused and recognition precision deteriorates
Solution Approach 1:
The patent extracts and removes redundant data through principal component analysis (PCA) to retain only the most informative features. By identifying and extracting the principal components that capture the essential variance in the data, the system eliminates over-fitting while maintaining training reliability.
Solution Approach 2:
The patent changes the parameter of data dimensionality by transforming the original high-dimensional signal data into a lower-dimensional space using PCA. This parameter transformation reduces the number of features while preserving the most significant information, thereby preventing over-fitting and improving recognition precision.
2Device complexity
If the feature extraction is simplified, then the system complexity is reduced, but if the feature is too single, imprecise recognition is caused
Solution Approach 1:
The patent applies parameter changes by transforming the signal data through PCA to create an optimized feature space. This transformation simplifies the feature extraction process while maintaining sufficient feature richness for precise recognition, as the principal components capture the essential characteristics of human activities.
3Reliability
If a large amount of data is processed for training, then more comprehensive training is achieved, but it takes a lot of time costs
Solution Approach 1:
The patent extracts only the most informative data components through PCA, eliminating redundant information that would increase processing time. By working with a reduced set of principal components, the system achieves comprehensive training with significantly lower time costs.
Solution Approach 2:
The patent changes the data representation parameters through PCA transformation, converting the original large-volume data into a compact form that retains essential information. This parameter transformation enables comprehensive training to be achieved with minimal processing time.
4Measurement precision
If the data is processed in detail, then the recognition precision is improved, but the processing time increases
Solution Approach 1:
The patent applies parameter changes by transforming the data into principal components that capture the most significant variance. This transformation enables detailed processing to be achieved efficiently, as the reduced-dimensional data requires less computational effort while maintaining high recognition precision.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach effectively denoises signals, reduces system complexity, and enables high-precision location-free sensing with minimal data, allowing for accurate recognition of human activities in various indoor settings.
Implementation Method 1
a radio frequency signal is propagated in a wireless medium through multiple paths, reflects different objects and reaches a receiver, so that the radio frequency signal carries environment-related information
Implementation Method 2
filtering the reflected signal by principal component analysis to obtain the noise-removed reflection signal
Implementation Method 3
A feature is extracted from a wireless signal in an indoor environment through the deep learning, so that the human activity may be recognized
Data Source
AI summary
Disclosed are an indoor non-contact human activity recognition method and system. The method comprises: collecting an indoor reflected signal by using an antenna array; filtering the reflected signal to obtain a noise-removed reflection signal; and inputting the noise-removed reflected signal to a pre-trained human activity recognition model, and determining a human activity category, the human activity recognition model being a pre-trained CNN network model based on a transfer learning algorithm. The recognition method and system have the advantages that: the antenna array is configured for collecting human actions to carry out activity recognition indoors, which can be applied to home-based care scenes; original data is denoised, so that most of high-frequency noises can be removed, and a phase change of the signal is reserved; a CNN structure is adopted for training so as to reduce a complexity of the system location-free sensing.