CN103792387B - A kind of recognition methods of the human body forward-reverse based on accelerometer - Google Patents
A kind of recognition methods of the human body forward-reverse based on accelerometer Download PDFInfo
- Publication number
- CN103792387B CN103792387B CN201410059809.1A CN201410059809A CN103792387B CN 103792387 B CN103792387 B CN 103792387B CN 201410059809 A CN201410059809 A CN 201410059809A CN 103792387 B CN103792387 B CN 103792387B
- Authority
- CN
- China
- Prior art keywords
- direction acceleration
- human body
- vertical direction
- peak value
- acceleration
- 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.)
- Active
Links
Landscapes
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The invention discloses a kind of recognition methods of the human body forward-reverse based on accelerometer, comprise the following steps: the vertical direction acceleration a of the human motion that (1) collects for accelerometer
zwith working direction acceleration a
y, carry out FIR low-pass filtering, remove high frequency noise; (2) detect the peak value of vertical direction acceleration, and preserve the sampling point position residing for peak value; (3) with vertical direction acceleration peak value place sampling point position, judge that the sampled point of the forward acceleration of the same position of synchronization is in trend or the downtrending of rising, if the ascendant trend of being in, then judge that human body is as retrogressing, otherwise show that human body is advance.
Description
Technical field
The present invention relates to a kind of method of object action Activity recognition, particularly relate to the recognition methods of a kind of human body forward-reverse action in indoor navigation location.
Background technology
Along with society and economic develop rapidly, the demand of people to position & navigation increases day by day, utilizes the information being arranged on the wireless senser collection of human body waist to carry out Activity recognition, is the important research direction of of indoor positioning.These behaviors comprise walking, stair activity, running etc., and wherein walking can be divided into walking (advance) forward, backward walking (retrogressing), side direction walking.
Backward movement is generally the motion of comparatively low speed, particularly in the on-the-spot such scene of fire-fighting, retrogressing situation at a high speed may be had hardly to occur.Move forward and backward action as shown in Figure 1, and in figure, the direction of arrow represents the direction of walking, and left figure advances, and right figure retreats.In pedestrian's island inner position, two the different actions that move forward and backward are consistent by the course that the sensor being fixed on human body waist obtains, and the actual motion direction of two actions is antipodal.So the identification of the action that moves forward and backward of pedestrian is very important.In other words, the identification moved forward and backward is related to the accuracy problem of indoor positioning.The geometric properties (cycle of extracting signal to the recognizer of backward movement in prior art, the crest of signal, trough), the energy feature etc. of signal, will carry out pre-service to signal data before extraction feature, not only calculated amount is large, algorithm is complicated, and discrimination is unsatisfactory.
The purposes of indoor positioning very extensive.Such as, the demand of indoor positioning day by day highlights in fire-fighting domain, under the situation that current particularly serious pernicious disaster, fire failure and chemical leakage accident are increasing, in the face of various types of severe disaster, how while effectively promoting disaster assistance progress and efficiency, First Line fireman individual safety can be ensured again to greatest extent, just becoming a realistic problem urgently to be resolved hurrily.Along with the progress of technology, the especially progress of location and navigation technology, provides the solution of science more by the solution for this problem.
Summary of the invention
Goal of the invention: technical matters to be solved by this invention is large for prior art calculated amount, that discrimination is undesirable deficiency, provides a kind of recognition methods of the human body forward-reverse based on accelerometer.
In order to solve the problems of the technologies described above, the invention discloses a kind of recognition methods of the human body forward-reverse based on accelerometer, comprising the following steps:
(1) the vertical direction acceleration a of human motion accelerometer collected
zwith working direction acceleration a
y, carry out FIR low-pass filtering, remove high frequency noise;
(2) detect the peak value of vertical direction acceleration, and preserve the sampling point position residing for peak value;
(3) with vertical direction acceleration peak value place sampling point position, judge that the sampled point of the working direction acceleration of the same position of synchronization is in trend or the downtrending of rising, if the ascendant trend of being in, then judge that human body is as backward movement, otherwise show that human body is forward motion.
The specific algorithm of step of the present invention (3) is:
(1) if, a
y(k+1) > a
y(k) > a
y(k-1), then judge that this action of human body is as retrogressing;
(2) if, a
y(k+1) < a
y(k) < a
y(k-1), then judge that this action of human body is as advance;
Wherein, a
yk () represents the point on the working direction Acceleration pulse corresponding with the sampled point of the crest of vertical direction acceleration, a
y(k-1) some a is represented
y(k) more front, a
y(k+1) some a is represented
y(k) rear.
Remove in step of the present invention (2) and be less than threshold value 10 ~ 12.5m/s
2the peak value of vertical direction acceleration.
Step of the present invention (2) if in mistiming between two vertical direction acceleration peak values be less than time threshold 0.1 ~ 0.3s, then a rear peak value is removed.
Retreat Activity recognition principle: the different actions considering human body, acceleration signal has different features, and these features comprise geometric properties (cycle, the crest of signal, trough), the energy feature etc. of signal.The present invention, by the geometric properties of identification two axle acceleration, distinguishes and moves forward and backward.
Beneficial effect: it is simple, effective that the present invention retreats Activity recognition method, by carrying out simple process to acceleration that is vertical and forward direction two axles, can judge retrogressing behavior, can reach following beneficial effect:
(1) calculated amount is little.Utilize the acceleration of vertical direction and working direction to carry out retrogressing Activity recognition, do not need the eigenwert such as average, variance calculating acceleration signal, reduce calculated amount
(2) identify completely.The data display of actual test, the discrimination that method for distinguishing is known in this retrogressing reaches 100%, can identify completely and move forward and backward, and the method is applicable to different people.
(3) real-time is good.To count based on step, normal acceleration often detects a peak value, is row and makes a move, and often row makes a move and can judge to advance or retreat.
Accompanying drawing explanation
To do the present invention below in conjunction with the drawings and specific embodiments and further illustrate, above-mentioned and/or otherwise advantage of the present invention will become apparent.
Fig. 1 is before and after human body and the posture figure retreated schemes.
Three direction of principal axis schematic diagram of Fig. 2 accelerometer.
Fig. 3 is process flow diagram of the present invention.
Fig. 4 is the Acceleration pulse figure of the forward direction that moves forward and backward of human body and vertical direction.
Embodiment
The invention discloses a kind of recognition methods of the human body forward-reverse based on accelerometer, comprise the following steps:
(1) the vertical direction acceleration a of human motion accelerometer collected
zwith working direction acceleration a
y, carry out FIR low-pass filtering, remove high frequency noise;
(2) detect the peak value of vertical direction acceleration, and preserve the sampling point position residing for peak value;
(3) with vertical direction acceleration peak value place sampling point position, judge that the sampled point of the working direction acceleration of the same position of synchronization is in trend or the downtrending of rising, if the ascendant trend of being in, then judge that human body is as retrogressing, otherwise show that human body is advance.
The specific algorithm of step of the present invention (3) is:
(1) if, a
y(k+1) > a
y(k) > a
y(k-1), then judge that this action of human body is as retrogressing;
(2) if, a
y(k+1) < a
y(k) < a
y(k-1), then judge that this action of human body is as advance;
Wherein, a
yk () represents the point on the working direction Acceleration pulse corresponding with the sampled point of the crest of vertical direction acceleration, a
y(k-1) some a is represented
y(k) more front, a
y(k+1) some a is represented
y(k) rear.
Remove in step of the present invention (2) and be less than threshold value 10 ~ 12.5m/s
2the peak value of vertical direction acceleration.
Step of the present invention (2) if in mistiming between two vertical direction acceleration peak values be less than time threshold 0.1 ~ 0.3s, then a rear peak value is removed.
Embodiment
The system of the inventive method application is Wearable dead reckoning navigation system, adopts MEMS (micro electro mechanical system) (MEMS) sensor, comprises three-axis gyroscope, three axis accelerometer and three axle magnetometers.Estimate step-length based on pedestrian's gait feature, associating course information calculates the position of pedestrian.
The present invention utilizes the signal of accelerometer, carries out retrogressing Activity recognition.As shown in Figure 2, sensor is fixed on human body waist, places according to the direction of presetting, and three axis accelerometer is made up of three orthogonal single-axis accelerometers, then the direction that three Acceleration pulses of sensor collection are corresponding is respectively known.Accelerometer three is axial to be defined as, a
zrepresent the acceleration of vertical direction, contrary with the direction of acceleration of gravity, a
yrepresent the acceleration of human body working direction, a
xrepresent lateral acceleration, with a
y, a
zform right-handed coordinate system.In the present invention, only a is needed
y, a
zthe acceleration signal of diaxon.
As shown in Figure 3, the step of the present embodiment is as follows:
(1) respectively FIR low-pass filtering is carried out to vertical direction acceleration and working direction acceleration, remove high frequency noise.
(2) peakvalue's checking is carried out to vertical direction acceleration, and preserve the sampling point position residing for peak value.
(3) with vertical direction acceleration peak value place sampling point position, judge that the sampled point of the working direction acceleration of the same position of synchronization is in trend or the downtrending of rising, if the ascendant trend of being in, then judge that human body is as retrogressing, otherwise show that human body is advance.When carrying out indoor positioning, need each step that will move forward and backward, the correspondence counter that moves forward and backward adds 1 process respectively.
The Acceleration pulse figure of the forward direction of action and the vertical direction of moving forward and backward as shown in Figure 4.
Can obviously find out from Fig. 4, when vertical direction acceleration is in crest value, the state corresponding to working direction acceleration: during advance, working direction acceleration is in downtrending; During retrogressing, working direction acceleration is in ascendant trend.
Judge rise and fall trend by the size of more a certain sampled point and last sampled point and a rear sampled point, namely have the value of a sampled point larger than a upper sampled point, less than next sampled point, then this point is in ascendant trend; On the contrary, then this point is in downtrending.Thus can judge to retreat or advance according to the rise and fall trend of working direction acceleration corresponding to vertical direction acceleration peak value point.Arthmetic statement is:
(1) a
y(k) > a
y, and a (k-1)
y(k) < a
y(k+1), then this action is retrogressing;
(2) a
y(k) < a
y, and a (k-1)
y(k) > a
y(k+1), then this action is advance;
The walking movement of human body is not absolute periodic, therefore, directly carries out peakvalue's checking to acceleration, can cause flase drop.Employing FIR (FiniteImpulseResponse) low-pass filtering carries out peakvalue's checking after carrying out filtering to acceleration, improves accuracy in detection.
FIR filtering is carried out to vertical direction acceleration, chooses moving window, the sample maximum in real-time judge window as capturing a potential acceleration peak value, the sliding window of continuous moving, the final peak value selecting acceleration in a step.The present embodiment, in order to improve the accuracy rate of peakvalue's checking, adds following two constraint conditions when peakvalue's checking:
(1) acceleration peak value must be greater than threshold value 10.5m/s
2;
Mistiming between (2) two peak values must be greater than time threshold 0.2s.
The invention provides a kind of recognition methods of the human body forward-reverse based on accelerometer; the method and access of this technical scheme of specific implementation is a lot; the above is only the preferred embodiment of the present invention; should be understood that; for those skilled in the art; under the premise without departing from the principles of the invention, can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.The all available prior art of each ingredient not clear and definite in the present embodiment is realized.
Claims (1)
1., based on a recognition methods for the human body forward-reverse of accelerometer, it is characterized in that, comprise the following steps:
(1) the vertical direction acceleration a of human motion accelerometer collected
zwith working direction acceleration a
y, carry out FIR low-pass filtering, remove high frequency noise;
(2) detect the peak value of vertical direction acceleration, and preserve the sampling point position residing for peak value;
(3) with vertical direction acceleration peak value place sampling point position, judge that the sampled point of the working direction acceleration of the same position of synchronization is in trend or the downtrending of rising, if the ascendant trend of being in, then judge that human body is as backward movement, otherwise show that human body is forward motion;
The specific algorithm of step (3) is:
(1) if, a
y(k+1) > a
y(k) > a
y(k-1), then judge that this action of human body is as retrogressing;
(2) if, a
y(k+1) < a
y(k) < a
y(k-1), then judge that this action of human body is as advance;
Wherein, a
yk () represents the point on the working direction Acceleration pulse corresponding with the sampled point of the crest of vertical direction acceleration, a
y(k-1) some a is represented
y(k) more front, a
y(k+1) some a is represented
y(k) rear;
Remove in step (2) and be less than threshold value 10 ~ 12.5m/s
2the peak value of vertical direction acceleration;
Step (2) if in mistiming between two vertical direction acceleration peak values be less than time threshold 0.1 ~ 0.3s, then a rear peak value is removed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410059809.1A CN103792387B (en) | 2014-02-21 | 2014-02-21 | A kind of recognition methods of the human body forward-reverse based on accelerometer |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410059809.1A CN103792387B (en) | 2014-02-21 | 2014-02-21 | A kind of recognition methods of the human body forward-reverse based on accelerometer |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103792387A CN103792387A (en) | 2014-05-14 |
| CN103792387B true CN103792387B (en) | 2016-01-20 |
Family
ID=50668255
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410059809.1A Active CN103792387B (en) | 2014-02-21 | 2014-02-21 | A kind of recognition methods of the human body forward-reverse based on accelerometer |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103792387B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105092891B (en) * | 2015-09-01 | 2019-10-01 | 深圳Tcl数字技术有限公司 | Terminal gets rid of screen recognition methods and device |
| CN109331389A (en) * | 2018-11-12 | 2019-02-15 | 重庆知遨科技有限公司 | A kind of fire-fighting robot movement method of real-time based on Multi-sensor Fusion |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103245797A (en) * | 2012-02-07 | 2013-08-14 | 精工爱普生株式会社 | State detection device, electronic apparatus, and program |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP3775779B2 (en) * | 2000-10-30 | 2006-05-17 | 株式会社国際電気通信基礎技術研究所 | Walking navigation device and navigation system using the same |
| JP4126388B2 (en) * | 2002-04-08 | 2008-07-30 | カシオ計算機株式会社 | Walking direction detecting device and program |
-
2014
- 2014-02-21 CN CN201410059809.1A patent/CN103792387B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103245797A (en) * | 2012-02-07 | 2013-08-14 | 精工爱普生株式会社 | State detection device, electronic apparatus, and program |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103792387A (en) | 2014-05-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6567658B2 (en) | Device and method for classifying user activity and / or counting user steps | |
| CN104406603B (en) | A kind of step-recording method and device based on acceleration transducer | |
| CN107343789B (en) | Step motion identification method based on three-axis acceleration sensor | |
| CN103886715B (en) | A kind of tumble detection method for human body | |
| CN104914468A (en) | Mine micro-quake signal P wave first arrival moment joint pickup method | |
| CN111089604B (en) | Body-building exercise identification method based on wearable sensor | |
| CN103792387B (en) | A kind of recognition methods of the human body forward-reverse based on accelerometer | |
| Yang et al. | A wearable real-time fall detector based on Naive Bayes classifier | |
| CN104931057B (en) | A kind of any position localization method, the apparatus and system of robot | |
| CN107303181B (en) | Step motion identification method based on six-axis sensor | |
| CN104567912A (en) | Method for realizing pedometer on Android mobile phone | |
| CN103637782A (en) | Human health monitoring system and method based on intelligent terminal | |
| CN105303183A (en) | Children's attitude identification and analysis system and method based on wearable equipment | |
| CN102799875A (en) | Tracing method of arbitrary hand-shaped human hand | |
| CN109540143A (en) | The unconventional direction of action recognition methods of pedestrian of more sensing sources dynamic peak value fusions | |
| Zhen et al. | Wearable preimpact fall detector using SVM | |
| CN109720353B (en) | Driving behavior detection method based on smart phone | |
| Ye et al. | Research of fall detection and alarm applications for the elderly | |
| JP2013088280A5 (en) | ||
| CN104605859B (en) | Indoor navigation gait detection method based on mobile terminal sensor | |
| CN103632133A (en) | Human gesture recognition method | |
| CN111626273B (en) | Fall behavior recognition system and method based on atomic action time sequence characteristics | |
| CN105551191A (en) | Falling detection method | |
| CN112556831B (en) | Detection method of intelligent detection alarm device for well lid abnormity | |
| CN104346606A (en) | Abnormal gait analyzing method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |