Search

CN-121995367-A - Household old-care privacy protection type activity detection method based on infrared and microwave radar cooperation

CN121995367ACN 121995367 ACN121995367 ACN 121995367ACN-121995367-A

Abstract

The invention belongs to the field of intelligent monitoring for home-based aged people, aims to solve the problems of privacy leakage, low efficiency in cooperation, single monitoring and the like in the prior art, and provides a wearable and nonimaging activity detection method. The method adopts a passive infrared sensor and a 24GHz non-imaging microwave radar to construct a cooperative system, and is matched with an edge computing and privacy enhancing module, and the anti-interference capability and low power consumption balance are improved through three-level hierarchical scheduling and scene dynamic weight fusion of 'infrared pre-triggering and microwave radar verification'; the method comprises the steps of constructing a signal layer non-imaging acquisition, processing layer local calculation and transmission layer encryption desensitization triple privacy protection, avoiding privacy disclosure, identifying 6 types of activity states through multi-feature combination, constructing a regular baseline by combining 7 days of data, and realizing 'real-time + regular deviation' double-dimension warning. The method does not need the cooperation of the old, reduces the false alarm rate by more than 40%, adapts to the requirements of various home scenes and the old, and has strong practicability and easy popularization.

Inventors

  • WANG PENG
  • YANG SHUHAN

Assignees

  • 沈阳双杰网络科技集团有限公司

Dates

Publication Date
20260508
Application Date
20260206

Claims (6)

  1. 1. The household old-care privacy protection type activity detection method based on the cooperation of infrared and microwave radar is characterized by comprising the following steps of: S1, system deployment and initialization, namely deploying a monitoring terminal integrating a passive infrared sensor, a non-imaging microwave radar sensor and a local edge computing unit in a home key area, calibrating sensor parameters and constructing a home environment feature library; S2, layering collaborative triggering monitoring, namely adopting a layering scheduling mechanism of 'infrared pre-triggering and microwave radar accurate verification' to realize the balance of low power consumption and high-precision monitoring; s3, activity state identification and rule modeling are carried out, multi-dimensional characteristics of the dual sensors are extracted, the activity state is identified through a lightweight neural network, and a personalized activity rule base line is constructed based on time sequence analysis; S4, abnormality judgment and privacy enhancement processing, namely identifying risk events through a two-dimensional abnormality judgment mechanism, and guaranteeing privacy safety through a full-link protection strategy; And S5, alarming and self-adaptive optimization, triggering hierarchical alarming and continuously optimizing system parameters and a regular base line based on new acquired data.
  2. 2. The home care privacy preserving activity detecting method based on infrared and microwave radar cooperation as claimed in claim 1, wherein the S1 comprises the following sub-steps: s11, deploying a home terminal, deploying a monitoring terminal in a bedroom, a bathroom, a living room and other home key active areas, wherein the passive infrared sensor only collects human body thermal radiation change signals, and the non-imaging microwave radar sensor shields a radio frequency imaging function and only extracts target motion characteristics and existence characteristics; S12, calibrating sensor parameters, namely calibrating a thermal radiation threshold range of the passive infrared sensor to be 8-12 mu W/cm < 2 > through the local edge computing unit, shielding environmental heat source interference through a multi-frame differential algorithm, calibrating a detection distance range of the non-imaging microwave radar sensor to be 0.5-5m, and calibrating a sampling frequency to be 10Hz, and eliminating electromagnetic interference through a space-frequency domain filtering algorithm; S13, constructing an environment feature library, inputting environment information such as furniture positions, fixed heat source distribution and the like of a home scene, and building an environment reference feature library.
  3. 3. The home-care privacy preserving activity detection method based on infrared and microwave radar coordination of claim 1, wherein S2 comprises the following sub-steps: s21, monitoring a sleep level, wherein when no effective infrared signal is triggered, the non-imaging microwave radar sensor is in a low-power-consumption sleep state, and only the passive infrared sensor monitors in real time with 2S as a period; S22, triggering level awakening, wherein when the passive infrared sensor detects that the thermal radiation signal variation exceeds a set threshold value and the duration is more than or equal to 0.5S, the non-imaging microwave radar sensor is awakened; S23, accurate monitoring level fusion, wherein the non-imaging microwave radar sensor is awakened and then synchronously acquires data with the passive infrared sensor, the local edge computing unit carries out dynamic weight fusion on the dual-source data, the infrared weight is 0.4 and the microwave radar weight is 0.6 in a dynamic scene, and the infrared weight is 0.2 and the microwave radar weight is 0.8 in a static scene.
  4. 4. The home-care privacy preserving activity detection method based on infrared and microwave radar coordination of claim 1, wherein S3 comprises the following sub-steps: s31, extracting multidimensional features, namely extracting heat radiation change rate and signal duration features from the passive infrared sensor, and extracting target distance change rate, motion amplitude and static duration features from the non-imaging microwave radar sensor; s32, classifying lightweight neural networks, and identifying 6 active states of sitting, standing, walking, standing/sitting, lying and abnormal postures by adopting MobileNetV lightweight convolutional neural network models; And S33, constructing a baseline by time sequence analysis, continuously collecting 7-day activity data, and constructing a personalized activity rule baseline by an ARIMA time sequence analysis algorithm.
  5. 5. The home-care privacy preserving activity detection method based on infrared and microwave radar coordination of claim 1, wherein S4 comprises the following sub-steps: S41, judging two-dimensional abnormality, wherein the real-time state abnormality is that a first-level alarm is triggered when dangerous postures such as falling and curling are detected, the rule deviation abnormality is that a second-level alarm is triggered when the deviation between the current active state and a base line exceeds a set threshold value, and the second-level alarm is verified by the non-imaging microwave radar sensor for the second time before the second-level alarm; S42, full-link privacy protection, wherein privacy zero leakage is ensured through a triple protection strategy, ①, namely, a signal layer, a sensor only collects non-imaging original signals and does not generate any privacy data such as images, body states and the like, ②, namely, all data are processed by a local edge computing unit and are not transmitted to the original signals, state results (such as 'normal bedridden' and 'falling alarm') are only uploaded, ③, namely, a TLS 1.3 encryption transmission protocol is adopted for alarm information and state data, and a differential privacy technology is matched for data desensitization, so that any characteristic information of an old person capable of being correlated with the old person is deleted.
  6. 6. The home-care privacy preserving activity detection method based on infrared and microwave radar coordination of claim 1, wherein S5 comprises the following sub-steps: S51, triggering a hierarchical alarm, wherein the primary alarm is synchronously triggered by local acousto-optic prompt, guardian mobile phone APP pushing and emergency contact short message, and the secondary alarm only pushes guardian mobile phone APP; and S52, optimizing the self-adaptive parameters, and automatically optimizing the regular base line and the sensor parameters based on the new acquired data every 30 days by the system.

Description

Household old-care privacy protection type activity detection method based on infrared and microwave radar cooperation Technical Field The invention relates to the technical field of intelligent monitoring of home-based aged people, in particular to a home-based aged people privacy protection type activity detection method based on cooperation of infrared and microwave radars, which is suitable for accurately monitoring daily activity states of aged people in a wearable and non-imaging scene and simultaneously realizes privacy zero-leakage protection. Background With the aggravation of the aging society, home-based care is a mainstream care mode, and the monitoring of the daily activity state of the old is a core requirement for guaranteeing care safety. The existing home-based aged monitoring technology is mainly divided into two types of wearable monitoring and non-wearable monitoring, wherein wearable equipment is required to be actively worn by the old, the problems of poor compliance, easy falling, incapacity of adapting to the old and the like exist, and in the non-wearable monitoring, the risk of serious privacy leakage exists in the monitoring of a camera, and the monitoring is difficult to be accepted by the old. In order to balance the monitoring requirement and privacy protection, the prior art adopts a scheme of combining millimeter wave radar and infrared thermal imaging, and realizes monitoring such as falling and heart rate through data complementation. However, the scheme still has obvious defects that the infrared thermal imaging is a weak imaging technology in nature, privacy information such as the state and behavior details of the old still can be revealed, the sensor cooperation mode is mostly parallel data superposition, scene self-adaptive scheduling is lacked, interference from home environments (such as furniture shielding and temperature fluctuation) is easy to occur, the false alarm rate is high, most schemes only focus abnormal event (falling and heart rate abnormality) detection, dynamic capture and trend prejudgment of the daily activity rule of the old cannot be realized, and the fine care requirement of home-based aged is difficult to meet. In addition, part of the prior art adopts a single infrared or microwave radar sensor, the single infrared sensor has weak identification capability to static old people and is easy to be subjected to 'missed monitoring', the single microwave radar sensor is easy to be interfered by electric appliances in a complex household environment, and the identification precision to actions with smaller activity amplitude (such as standing up and sitting down gesture change) is insufficient. Therefore, there is a need for a home care detection method that is completely non-imaging, sensor-collaborative and efficient, and has both privacy protection and multi-dimensional activity monitoring capabilities. Disclosure of Invention The invention aims to solve the defects of high privacy leakage risk, low sensor cooperative efficiency, single monitoring dimension, weak anti-interference capability and the like in the existing home-based care monitoring technology, and provides a home-based care privacy protection type activity detection method and system based on infrared and microwave radar cooperative, which realize accurate monitoring and pre-judging of daily activity states, behavior rules and abnormal events of the old on the premise of completely avoiding privacy leakage. In order to achieve the aim, the invention adopts the following technical scheme that the household care privacy protection type activity detection method based on the cooperation of infrared and microwave radars comprises the following steps: s1, system deployment and initialization are carried out, and a non-imaging collaborative monitoring hardware base and an environment reference feature library are constructed; S2, layering collaborative triggering monitoring, namely realizing the balance of low power consumption and high-precision monitoring through a layering scheduling mechanism of 'infrared pre-triggering and microwave radar precision verification'; S3, activity state identification and rule modeling are carried out, activity state classification is achieved based on double-source multidimensional feature fusion, and a personalized activity rule base line is established; S4, abnormality judgment and privacy enhancement processing, namely identifying risk events through a two-dimensional abnormality judgment mechanism, and guaranteeing privacy safety through a full-link protection strategy; and S5, alarming and self-adaptive optimization, triggering grading alarming and continuously optimizing system parameters and a regular base line. As a further scheme of the present invention, the step S1 specifically includes: And S11, deploying a home terminal, deploying a monitoring terminal in a bedroom, a bathroom, a living room and other home key active areas, wherein each termin