Intelligent human fall recognition and monitoring system
Patent Information
- Application Number
- TW115203295
- Authority / Receiving Office
- TW · TW
- Patent Type
- Utility models
- Current Assignee / Owner
- Filing Date
- 2026-04-16
- Publication Date
- 2026-09-01
- Estimated Expiration
- 2036-04-15
Smart Images
Figure TWG2TB001909378_001 
Figure TWG2TB001909378_002 
Figure TWG2TB001909378_003
Abstract
Claims
1. A smart human fall recognition and monitoring system, comprising: an image capturing module for capturing real-time images within a predetermined range and generating real-time image data; a face recognition module, communicatively connected to the image capturing module, for analyzing human images in the real-time image data using AI image processing technology and generating at least one real-time face data; and a posture recognition and analysis module, communicatively connected to the image recognition module, for analyzing the movements of the face data using AI image processing technology, determining that the face data belongs to one of a standing posture, a sitting posture, or a lying posture, and generating corresponding real-time posture data. A fall detection module, communicatively connected to the posture recognition and analysis module, is used to determine whether the time difference of posture change from standing or sitting to lying in the real-time posture data is less than a fall warning threshold; when the posture change time difference is less than the fall warning threshold, the fall detection module determines that at least one person in the real-time video data has fallen and issues a notification signal; and an alert module, communicatively connected to the fall detection module, is used to issue an alert signal after receiving the notification signal from the fall detection module.
2. The intelligent human fall recognition and monitoring system as described in claim 1, wherein the posture recognition and analysis module includes an aspect ratio analysis model, having an aspect ratio parameter measurement unit and an aspect ratio calculation unit; the aspect ratio parameter measurement unit is used to measure a height and a width of the human image data; the aspect ratio calculation unit is used to calculate an aspect ratio based on the height and width measured by the aspect ratio parameter measurement unit, wherein.
3. The intelligent human fall recognition and monitoring system as described in claim 2, wherein the aspect ratio analysis model further includes an aspect ratio judgment unit for comparing the aspect ratio with an upper limit value and a lower limit value; if the aspect ratio is positive, the aspect ratio judgment unit outputs a standing posture result; if the aspect ratio is negative, the aspect ratio judgment unit outputs a lying posture result; if the aspect ratio is negative, the aspect ratio judgment unit outputs a lying posture result.
4. The intelligent human fall recognition and monitoring system as described in claim 3, wherein the posture recognition and analysis module further includes a knee angle analysis model, having a knee angle parameter measurement unit and a knee angle calculation unit; the knee angle parameter measurement unit is used to measure a hip coordinate, a knee coordinate, an ankle coordinate, a first vector from the knee coordinate to the hip coordinate, and a second vector from the knee coordinate to the ankle coordinate of the human image data; the knee angle calculation unit is used to calculate a knee angle based on the first vector and the second vector measured by the knee angle parameter measurement unit, wherein.
5. The intelligent human fall detection and monitoring system as described in claim 4, wherein the hip coordinates, the knee coordinates, the ankle coordinates, the first vector, and a second vector from the knee coordinates to the ankle coordinates.
6. The intelligent human fall recognition and monitoring system as described in claim 4, wherein the knee angle calculation unit is used to calculate a left knee angle and a right knee angle respectively based on the left foot and right foot of the human image data, and to calculate an average knee angle based on the left knee angle and the right knee angle, wherein.
7. The intelligent human fall recognition and monitoring system as described in claim 6, wherein the knee angle analysis model further includes a knee angle judgment unit for comparing the relationship between the average knee angle and a knee angle constant; if the average knee angle is less than the knee angle, the knee angle judgment unit outputs the lying posture result; if the average knee angle is less than the knee angle, the knee angle judgment unit outputs the sitting posture result; and if the average knee angle is less than the knee angle, the knee angle judgment unit outputs the standing posture result.
8. The intelligent human fall recognition and monitoring system as described in claim 3, wherein the posture recognition and analysis module further includes a torso angle analysis model, having a torso angle parameter measurement unit and a torso angle calculation unit; the torso angle parameter measurement unit is used to measure a vertical vector and a torso vector of the human image data; the torso angle calculation unit is used to calculate a torso angle based on the vertical vector and the torso vector, wherein.
9. The intelligent human fall recognition and monitoring system as described in claim 8, wherein the torso angle analysis model further includes a torso angle judgment unit for comparing the relationship between the torso angle and a torso angle constant; if the relationship is positive, the torso angle judgment unit outputs the lying posture result; if the relationship is negative, the torso angle judgment unit outputs the sitting posture result.
10. The intelligent human fall recognition and monitoring system as described in claim 1 further includes a display module, which is communicatively connected to the image acquisition module, the posture recognition and analysis module and the fall detection module; the display module has a plurality of display blocks for displaying at least one of a real-time image, a posture statistics screen and a fall detection screen respectively.
11. The intelligent human fall recognition and monitoring system as described in claim 10 further includes a skeleton image conversion module, which is communicatively connected to the image capturing module and the display module, for converting the portraits of each person in the real-time image data into skeleton images through AI image processing technology, and thereby generating skeleton image data to be displayed on the real-time image screen of the display module.
12. The intelligent human fall detection and monitoring system as described in claim 1, wherein after issuing the notification signal, the fall detection module determines whether the time difference between the real-time posture data changing from the lying posture to the standing posture or the sitting posture is greater than the standing warning threshold; when the time difference between the second posture changes is greater than the standing warning threshold, the fall detection module issues a second notification signal to the warning module, so that the warning module issues a second warning signal after receiving the second notification signal from the fall detection module.