Radar Posture Recognition via Segmented Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current radar-based object detection systems face challenges in precise posture recognition due to environmental noise and high computational complexity, leading to low precision and high calculation demands in complex scenarios.
Innovation Solution
A radar-based posture recognition apparatus and method that combines spatial morphological and motion feature information from radar echo signals to determine postures such as lying on the floor, lying in bed, or walking, using a feature set that includes transient and process features, and satisfies specific condition sets to recognize multiple daily actions with high precision at a low computational cost.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If radar echo signals are analyzed to obtain target reflection points and cluster them to get position information, then the basic object detection function is achieved, but the precision of posture recognition is insufficient in complex environments
Solution Approach 1:
The patent segments the posture recognition process into multiple distinct feature extraction modules: spatial morphological features (target position, size, shape), motion features (velocity, acceleration, motion trajectory), and temporal features (feature changes over time). Each module processes specific aspects of the radar data independently, then combines them for comprehensive posture recognition, thereby improving precision without excessive system complexity
Solution Approach 2:
The patent changes multiple parameters simultaneously to improve posture recognition precision: it extracts and analyzes multiple feature parameters (spatial, motion, temporal), adjusts clustering parameters for target point grouping, and uses threshold parameters for posture classification. This multi-parameter approach enables accurate distinction between different postures (lying, sitting, standing, walking) in complex environments
2Measurement precision
If complex feature analysis is performed to improve posture recognition precision, then recognition accuracy improves, but computational complexity and calculation demands increase
Solution Approach 1:
The patent divides computational tasks into separate processing stages: first extracting spatial morphological features from clustered target points, then calculating motion features from temporal changes, and finally determining posture based on combined features. This segmentation allows efficient use of computational resources at each stage rather than performing all calculations simultaneously, reducing overall power consumption while maintaining high precision
Solution Approach 2:
The patent performs preliminary clustering of target reflection points to form target point sets before detailed feature analysis. This preliminary action reduces the data volume that requires complex processing by grouping related points together, thereby reducing computational power requirements for subsequent posture recognition while preserving 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
Enables accurate recognition of postures and daily actions with high precision and low computational requirements, improving the precision of posture recognition in complex environments.
Implementation Method 1
A radar emits electromagnetic waves via a transmit antenna, receives corresponding reflected waves after being reflected by different objects, and analyzes the received signals
Data Source
AI summary
Embodiments of this disclosure provide a radar-based posture recognition apparatus, method and an electronic device. The method includes: acquiring radar reflection point information based on radar echo signals reflected from a detected target, and clustering radar reflection points; in a first time period, calculating spatial morphological feature information and/or motion feature information of a target point set obtained by clustering; in a second time period, counting the spatial morphological feature information and/or the motion feature information of a plurality of first time periods to obtain motion process feature information in the second time period; and taking the motion process feature information within a second time period in which a current moment is present and the spatial morphological feature information and/or the motion feature information in a first time period in which the current moment is present as a feature set, to determine the posture of the detected target.