CN110473382A - A kind of fall detection algorithm - Google Patents
A kind of fall detection algorithm Download PDFInfo
- Publication number
- CN110473382A CN110473382A CN201910702414.1A CN201910702414A CN110473382A CN 110473382 A CN110473382 A CN 110473382A CN 201910702414 A CN201910702414 A CN 201910702414A CN 110473382 A CN110473382 A CN 110473382A
- Authority
- CN
- China
- Prior art keywords
- value
- data
- angle
- svm
- group
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/02—Alarms for ensuring the safety of persons
- G08B21/04—Alarms for ensuring the safety of persons responsive to non-activity, e.g. of elderly persons
- G08B21/0407—Alarms for ensuring the safety of persons responsive to non-activity, e.g. of elderly persons based on behaviour analysis
- G08B21/043—Alarms for ensuring the safety of persons responsive to non-activity, e.g. of elderly persons based on behaviour analysis detecting an emergency event, e.g. a fall
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/02—Alarms for ensuring the safety of persons
- G08B21/04—Alarms for ensuring the safety of persons responsive to non-activity, e.g. of elderly persons
- G08B21/0438—Sensor means for detecting
- G08B21/0446—Sensor means for detecting worn on the body to detect changes of posture, e.g. a fall, inclination, acceleration, gait
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Gerontology & Geriatric Medicine (AREA)
- Business, Economics & Management (AREA)
- Emergency Management (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Psychology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Emergency Alarm Devices (AREA)
- Alarm Systems (AREA)
Abstract
The present invention provides a kind of fall detection algorithms, comprising the following steps: pre-processes every group of data of acquisition, removes error amount；The SVM value of every group of data is calculated, and with the form calculus CV value of time window；If the SVM value of every group of data is greater than SVM threshold value, and CV value is all larger than CV threshold value in continuous three time windows, and angle is all larger than threshold angle in continuous three time windows, is judged as tumble.
Description
Technical field
The present invention relates to a kind of data identification technology, especially a kind of fall detection algorithm.
Background technique
In China, unexpected injury occupies the 4th in the common cause of the death of the elderly, and falling is to cause the elderly serious
The main reason for injury.Tumble also will cause the elderly's disability, psychological shade and the financial burden of family etc..Therefore, it falls and examines
Examining system is particularly important, and detection and alarm promptly and accurately can shorten therapeutic time, can be effectively reduced tumble and be brought
Unexpected injury, the psychological burden of the elderly is reduced, to promote the quality of life of the elderly.
Disclose a kind of fall detection method application No. is the patent of " 201510399133.5 ", including off-line training and
Line detects two stages.Multiple sons point are trained with random assortment Attributions selection in the selection of off-line training step combination random sample
Class device；Online fall detection stage, the classification results based on multiple sub-classifiers provide final judgement result.But pass through data
Combing, the threshold decision of single attribute is difficult accurately to judge tumble behavior in this application.
Summary of the invention
The purpose of the present invention is to provide a kind of fall detection algorithms.
Realize the technical solution of the object of the invention are as follows: a kind of fall detection algorithm, comprising the following steps:
Every group of data of acquisition are pre-processed, error amount is removed；
The SVM value of every group of data is calculated, and with the form calculus CV value of time window；
If the SVM value of every group of data is greater than SVM threshold value, and CV value is all larger than CV threshold value in continuous three time windows, and even
Angle is all larger than threshold angle in continuous three time windows, is judged as tumble.
Further,Wherein ax、ay、azThe acceleration in respectively three directions.
Further, the CV value calculation in a time window is as follows:
Wherein, SD is the standard deviation of the resultant acceleration in a time window.
Further, the angle W calculation method in a time window is as follows:
Wherein, wx、wyRespectively both direction angle.
Compared with prior art, the present invention it is a kind of based on resultant acceleration, acceleration to have the advantage that the present invention devises
Dispersion degree, the multistage of angle change judge algorithm, judge more acurrate.
The invention will be further described with reference to the accompanying drawings of the specification.
Detailed description of the invention
Fig. 1 is the method for the present invention flow diagram.
Specific embodiment
Before judgement is fallen, data sampling is carried out, the present embodiment chooses volunteer's carry sensors mould of different building shape
Quasi- Falls in Old People process, sensor collect the angle of 3-axis acceleration and both direction during all kinds of daily behaviors.
In order to avoid single experiment bring error, our every kind of daily behavior has done 100 groups of data statistics, and tumble behavior has done 300
Group data statistics, the elderly's daily routines and tumble behavior is divided into following several
| Serial number | Type of action | Specific movement |
| 1 | ADL | It stands |
| 2 | ADL | It sits down-stands up |
| 3 | ADL | It squats down-stands up |
| 4 | ADL | On foot |
| 5 | ADL | It jogs |
| 6 | ADL | It lies down |
| 7 | Fall | It falls forward |
| 8 | Fall | It falls back |
| 9 | Fall | It falls to side |
| 10 | ADL-Fall | Crouching-tumble |
| 11 | ADL-Fall | Race-tumble |
| 12 | ADL-Fall | It walks-falls |
| 13 | ADL-Fall | Seat-tumble |
Sensor and the consistent rectangular coordinate system in space in human body standing direction are initially set up.Gravity direction is Z axis, human body
Towards direction be X-axis, be Y-axis on the right side of human body.Then three directional accelerations of sensor acquisition are respectively as follows: ax、ay、az.Two
Angle direction is respectively that human body leans forward direction wxWith the direction w turned righty。
Collected data imported into for statistical analysis in MATLAB software by we, after removing error information, obtain
To such as drawing a conclusion
By the processing discovery to data, the threshold decision of single attribute is difficult accurately to judge tumble behavior.Therefore originally
Embodiment devise it is a kind of based on resultant acceleration, acceleration dispersion degree, angle change multistage judge algorithm.In conjunction with Fig. 1, originally
A kind of fall detection algorithm that embodiment is related to, comprising the following steps:
Step 1, every group of data of acquisition are subjected to pretreatment removal error amount；
Step 2, the SVM value of every group of data is calculated；
Step 3, if the SVM value of every group of data is not more than SVM threshold value, it is judged as normal activity；If the SVM value of every group of data
Greater than SVM threshold value, 4 are gone to step；
Step 4, the CV value in a time window is calculated, if the CV value of certain amount group data is not more than CV threshold in time window
Value, is judged as of short duration strenuous exercise；Otherwise, the CV value for taking certain amount group data in continuous three time windows, goes to step 5；
Step 5, if the CV value of certain amount group data is judged as not all greater than CV threshold value in continuous three time windows
Of short duration strenuous exercise；If the CV value of certain amount group data is all larger than CV threshold value in continuous three time windows, 6 are gone to step；
Step 6, if angle is judged as of short duration strenuous exercise not all greater than angle threshold in continuous three time windows；
If angle is all larger than angle threshold in continuous three time windows, it is judged as tumble.
The general used time 0.5s of people's tumble process known to data acquisition, therefore time window T is set as 15 sampled points.
It is three directional acceleration vector sums that SVM, which is resultant acceleration, in step 2, is obtained by formula (1):
Wherein ax、ay、azThe acceleration in respectively three directions.
CV value is the index of variability of normalization measurement in step 4, indicates the dispersion degree of resultant acceleration, is obtained by formula (2)
Wherein, SD is the standard deviation of the resultant acceleration in a time window.
Angle is indicated by both direction angle vector sum in step 6, such as formula (3)
Wherein, wx、wyRespectively both direction angle.
Embodiment one
Initial threshold is provided with by inspection information and experience first, preferential reduction rate of failing to report is then based on, drops as far as possible
The principle of low rate of false alarm gradually adjusts threshold value by the judgement to real data and result.By the verifying under nearly thousand groups of data
With debugging, we determined that such as lower threshold value:
SVM=1.65g CV=2.0 W=45 degree
Algorithm is inputted by testing adjusted threshold value, obtains final algorithm routine.It is living to resurvey 300 groups of the elderlys
Dynamic data carry out precision test.The accuracy rate for obtaining algorithm under current threshold value is up to 96%, and wherein rate of failing to report is reduced to 0.5%,
Rate of false alarm 3.5%.
Claims (4)
1. a kind of fall detection algorithm, which comprises the following steps:
Every group of data of acquisition are subjected to pretreatment removal error amount；
The SVM value of every group of data is calculated, and with the form calculus CV value of time window；
If the SVM value of every group of data is greater than SVM threshold value, and CV value is all larger than CV threshold value, and continuous three in continuous three time windows
Angle is all larger than threshold angle in a time window, is judged as tumble.
2. algorithm according to claim 1, which is characterized in thatWherein ax、ay、azRespectively
The acceleration in three directions.
3. algorithm according to claim 1, which is characterized in that the CV value calculation in a time window is as follows:
Wherein, SD is the standard deviation of the resultant acceleration in a time window.
4. algorithm according to claim 1, which is characterized in that the angle W calculation method in a time window is as follows:
Wherein, wx、wyRespectively both direction angle.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910702414.1A CN110473382A (en) | 2019-07-31 | 2019-07-31 | A kind of fall detection algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910702414.1A CN110473382A (en) | 2019-07-31 | 2019-07-31 | A kind of fall detection algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110473382A true CN110473382A (en) | 2019-11-19 |
Family
ID=68509534
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910702414.1A Pending CN110473382A (en) | 2019-07-31 | 2019-07-31 | A kind of fall detection algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110473382A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111210595A (en) * | 2020-01-15 | 2020-05-29 | 广东工业大学 | Human body falling detection and warning method, device and computer readable storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102227189A (en) * | 2008-11-28 | 2011-10-26 | 皇家飞利浦电子股份有限公司 | Method and apparatus for fall detection |
| CN103337132A (en) * | 2013-06-08 | 2013-10-02 | 山东师范大学 | Tumble detection method for human body based on three-axis acceleration sensor |
| US20150213702A1 (en) * | 2014-01-27 | 2015-07-30 | Atlas5D, Inc. | Method and system for behavior detection |
| CN108109336A (en) * | 2017-11-28 | 2018-06-01 | 北京品驰医疗设备有限公司 | A kind of human body tumble recognition methods based on acceleration transducer |
| CN109670396A (en) * | 2018-11-06 | 2019-04-23 | 华南理工大学 | A kind of interior Falls Among Old People detection method |
-
2019
- 2019-07-31 CN CN201910702414.1A patent/CN110473382A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102227189A (en) * | 2008-11-28 | 2011-10-26 | 皇家飞利浦电子股份有限公司 | Method and apparatus for fall detection |
| CN103337132A (en) * | 2013-06-08 | 2013-10-02 | 山东师范大学 | Tumble detection method for human body based on three-axis acceleration sensor |
| US20150213702A1 (en) * | 2014-01-27 | 2015-07-30 | Atlas5D, Inc. | Method and system for behavior detection |
| CN108109336A (en) * | 2017-11-28 | 2018-06-01 | 北京品驰医疗设备有限公司 | A kind of human body tumble recognition methods based on acceleration transducer |
| CN109670396A (en) * | 2018-11-06 | 2019-04-23 | 华南理工大学 | A kind of interior Falls Among Old People detection method |
Non-Patent Citations (1)
| Title |
|---|
| 郭元新等: "基于 MPU6050 传感器的跌倒检测算法", 《湖南工业大学学报》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111210595A (en) * | 2020-01-15 | 2020-05-29 | 广东工业大学 | Human body falling detection and warning method, device and computer readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109645990B (en) | Computer mode identification method for electroencephalogram signals of epileptics | |
| CN105023022B (en) | Fall detection method and system | |
| CN110367991B (en) | Old people falling risk assessment method | |
| CN104484644B (en) | A kind of gesture identification method and device | |
| CN109117730A (en) | Electrocardiogram auricular fibrillation real-time judge method, apparatus, system and storage medium | |
| Jensen et al. | Classification of kinematic swimming data with emphasis on resource consumption | |
| WO2017101529A1 (en) | Electrocardio lead intelligent selection method and system | |
| CN111288986B (en) | Motion recognition method and motion recognition device | |
| CN108021888B (en) | Fall detection method | |
| KR20120052610A (en) | Apparatus and method for recognizing motion using neural network learning algorithm | |
| CN108538388B (en) | Knee joint dyskinesia function judging method | |
| CN108109336A (en) | A kind of human body tumble recognition methods based on acceleration transducer | |
| CN107092861A (en) | Lower limb movement recognition methods based on pressure and acceleration transducer | |
| CN108717548B (en) | Behavior recognition model updating method and system for dynamic increase of sensors | |
| CN108875308B (en) | Method for judging and detecting pure phase proximity of dendrobium nobile lindl | |
| CN110473382A (en) | A kind of fall detection algorithm | |
| CN106901689A (en) | Cognitive and motor function detecting system and data processing method based on interaction in kind | |
| Youn et al. | Wearable sensor-based biometric gait classification algorithm using WEKA | |
| CN106227999B (en) | A kind of high-adaptability tumble behavioral value method | |
| Al-Mulla et al. | Statistical class separation using sEMG features towards automated muscle fatigue detection and prediction | |
| CN109598309A (en) | A kind of detection system and its monitoring method of mental package press machine | |
| CN117562311A (en) | Detection system of high-performance electronic cigarette atomizer | |
| CN112155520A (en) | Monitoring device and method based on body temperature data, electronic equipment and storage medium | |
| CN109598275A (en) | Feature selecting device, method and electronic equipment | |
| CN111428690A (en) | Identity authentication method based on gait signal topology analysis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20191119 |
|
| WD01 | Invention patent application deemed withdrawn after publication |