International Journal of Scientific & Engineering Research, Volume 6, Issue 4, April-2015 1209

ISSN 2229-5518

Activity Recognition Using Filtering Algorithm Varun Sharma (Member IEEE), Dean. Rajneesh Narula

Abstract— This paper inspect the trouble in recognition of exercise and also discover the burned calories after exercise. Our topic and motive of research to detect a problem, how walking, jogging and running effect on the human body using personalize algorithms. Pedometer is a device which is used to count the step while walking. This device may be used as a hardware device which attached with the body or may hold on body like in hand and count the steps. Every person have different way of walking and also difference in step distance, an informal calibration, performed by the user. An accelerometer is a device that measures proper acceleration. It is not necessary that accelerometer always measure a proper acceleration for each coordinate acceleration But the benefit of this research, it remotely observe the moves, calories and find accurate results in jogging and running. There are specific algorithm used f or finding the difference between the jogging and running, and detect the burned calories. W hole research performed in MATLAB platform. This platform is very reliable and flexible to use.

Index Terms— Low pass filter, high pass filter, Threshold, walking, jogging, running MATLAB.

—————————— ——————————

1 INTRODUCTION

HE activity reorganization algorithms working on a de- vice that is called as pedometer. A pedometer is a device, generally transferable and electronic or electromechanical,
that calculate each step taken by detecting the motion of the body hips [1]. As the analysis, walking is very useful exercise to maintain body. It also helpful to get the information and readings about the moves and burned calories. The whole rec- ognize will working on the three activities. This thesis is the process to filter the accurate results in walking, jogging and running. There are number of applications are available on the mobile app. stores. We are using personalize two algorithms on the place of accelerometer and gyroscope to obtain the ac- curate results in jogging and running. This concept widely used in exercise machines.
To SUB OBJECTIVE: The sub-objective of this project are
• To implement this concept in Matlab.
• To create a algorithm for the filters and another algo-
rithm for detecting exercise.

2 REVIEW STAGE

2.1 RELATED WORK

In this research, we discover that every exercise need a ex- act limitation that is also called as working threshold. Pedome- ter is a device which is used to count the step. There are a lot

————————————————

Varun Sharma pursuing.Tech in Information Technology at Adesh Insti- tute of Engineering and Technology ,Faridkot,Punjab, India Mob:-+91

98761-09948 E-mail:- varun.sharma9948@gmail.com

Dean. Rajneesh Narula Adesh Institute of Engineering and Technology

Faridkot, India.Mob:- +91 94170-07872 E-Mail:- raj_narula74@yahoo.com
of algorithms and working platform are available to describe the pedometer. Some related working ways are described bel- low:-
In a research work, the Nokia create a wrist–attached sen- sor platform which developed at the nokia research Center during one project to make possible research, show of utilize for wearable and wireless sensors. This wearable pedometer application was applied as one of the display of the capability of the platform. The step counting algorithm is describe and the performance is assess. The platform is design for running exercise. However, the step detection during walking is also discussed. 3-axis accelerometer and step counting algorithm are used for linear acceleration data. Step counting with help of wrist is an motivating topic. These exercises shows that it is possible but with the lot of challenges. This is functional step counter. improvements are probable and they have been well- known while working.
Human action or movement becomes an necessary factor of many applications for computing. In this paper describe about the iLearn system for classify a body activities using the Apple iPhone‟s 3-axis accelerometer and the Nike + iPod game Kit. Results propose activities as well as running, walking, bicy- cling, and sitting can be recognized without any guidance by an end-user.

A. Problem in similar exercise detection

This research work was conducted to statistically recognize the walking, jogging and running activities using the mobile sensors. With the appropriate application for exercise, moni- toring energy consumption of calorie compensation. With the help of a method this application get the each activity for analysis. To recognize all exercise data it accelerate data were continuously recoded from the subject performing each activi- ty from 50 m. in distance and take 20 seconds to show the giv- en results in statically way.

IJSER © 2015 http://www.ijser.org

International Journal of Scientific & Engineering Research, Volume 6, Issue 4, April-2015 1210

ISSN 2229-5518

Be owned the whole work its method is not able to show dis- tinguish between jogging and running.[3]

B. Head detection problem

This is another algorithm, it shows the working of step detec- tion and accurate working of pedometer. It get accuracy of the displacements measured at the step level cumulatively impact on the result.[8]

In an another research which based on the to acquire accurate results for head and step detection. Smartphone sensors are used with specific algorithm for receiving the input. A person hold the mobile in hand or pocket and move then its accelera- tion take input from the mobile sensors. Three operations are performed in this paper. Distance estimation, user heading interface and particle filtering. This whole process is used to set the proper output for head and step detection. There are some personal algorithms are used in these paper that are not declared yet. But there is major cause in this technique it get lot of errors while head detection. The head detection and magnetic interface are the major problems.[4]

Step Detection:- This algorithm is robust to random per- forming motions, Like- walking or not.

Stride Length Estimation:- It provide estimation for each

step or changing in position.

Heading Inference:- It detect the head position.

End-To-End System:- It shows the indoor positions using

smart-phone.
Start
Input Acc.samples
Compute Min & Max value
Acc>threshold
No
Compute threshold value

Acc within time

Window

Step Detection
End

Fig. Head detection problem

Fig : The overall system architecture

C. Problem formulation

Define Identifying the specific threshold where there pervious research not performing accurate working.
• Analyzing that how those reading from accelerometer and gyroscope are being misinterpreted.
• Developing an algorithm in MATLAB for improving the performance of the moves by improving its decision mak- ing.
• For detecting the proper signal 1st create algorithm for on low pass filter and high pass filter that filters works on the place of accelerometer and gyroscope.
• Calculating the No. of calories burnt by someone person with increased accuracy.
• There may be difference in men and women Calories burned process or algorithms.

CONCLUSION

There are lot of algorithms use here for exercise reorganiza- tion. The whole study about the working of pedometer use specific personalize algorithms to specify the accurate output. Like filtering algorithm [4] and machine learning algorithm [6] are effectively working on the testing but there may not get accurate results or there may be some problem in detection. After working with these algorithm we create a new algorithm which is very helpful to get accurate results. Some working ways are (1) Create a signal (input), (2) Filter the input accord- ing to the algorithm,(3) High Pass Filter and Low Pass Filter (4) Compare input signal (5) Recognize the final output/ re-

IJSER © 2015 http://www.ijser.org

International Journal of Scientific & Engineering Research, Volume 6, Issue 4, April-2015 1211

ISSN 2229-5518

sult. When we perform different experiments then get new results.

ACKNOWLEDGMENT

This study was conduct by the first author under the direction of the co-author in fractional accomplishment of the necessi- ties of a Master Degree in Information Technology. The First author wishes to thank Assistant Professor Harinder Singh at Adesh Institute of Engineering and Technology, Faridkot un- der Punjab Technical University , Jalandhar for his support over the period in which this article was written.

REFERENCES

[1] Vanda Ho, Rebecca K.Simmons,Charlotte L.Ridway,Esther M.F.van sluijs, Diane J.Bamber,lan M.Goodyer,Valerie J.Dunn, Ulf Ekelund,KristenCorder,"Is wearing a pedometer associated with higher physical activity among adolescents?", Elsevier Inc.2013.

[2] Catrine Tudor-Locke1 and David R. Bassett Jr2, "How Many

Steps/Day Are Enough?" Sports Med, 2004.

[3] Ekachai Thammasat," The Statistical Recognition of Walking, Jog- ging, and Running Using Smartphone Accelerometers",IEEE,2013.

[4] Fan Li," A Reliable and Accurate Indoor Localization Method Using

Phone Inertial Sensors",Sep-2012.

[5] Alvina Anjum and Muhammad U. Ilyas," Activity Recognition Using

Smartphone Sensors", IEEE,july-2013.

[6] T.Scott Saponas,Jonathan Lester, Jon Froehlich,James Forgarty, James Landay," iLearn on the iPhone: Real-Time Human Activity Classifica- tion on Commodity Mobile Phones.

[7] Koichi Ishida," Insole Pedometer With Piezoelectric Energy Harvest- er and 2 V Organic Circuits,"IEEE,2012.

[8] Kashyap Tumkur, Suneeth Subbiah," Modeling Human Walking for Step Detection and Stride Determination by 3-Axis Accelerometer Readings in Pedometer,IEEE-2012.

[9] Tom Mikael Ahola," Pedometer for Running Activity Using Accel- erometer Sensors on the Wrist, Libertas Academica," Medical Equipment Insights 2010.

[10] Jennifer R. Kwapisz, Gary M. Weiss, Samuel A. Moore," Activity

Recognition using Cell Phone Accelerometers, Washington, DC,July

25, 2010.

[11] Eric Chalmersa, Jonathan Leb, Dulai Sukhdeepc, Joe Wattd, John Andersend, Edmond Lou," Inertial sensing algorithms for long-term foot angle monitoring for assessment of idiopathic toe-walking," 31

August 2013.

[12] Yundong Xuana, b, Zhan Zhaoa, Zhen Fanga, , , Fangmin Suna, b, Zhihong Xua, b, Lidong Dua, Shaohua Wua," Gait Cycle Recognition based on Wireless Inertial Sensor Networ," 10.1016/j.ieri.2013.11.008.

[13] Omar Aziza, c, , , Edward J. Parkc, Greg Morid, Stephen N. Robi- novitcha, b, c," Distinguishing the causes of falls in humans using an array of wearable tri-axial accelerometers, 10.1016/j.gaitpost.2013.

08.034.

[14] Joan-Bryce Burla, , Anic Ostertag, Heike Schulze Westerath, Edna Hillmann," Gait determination and activity measurement in horses using an accelerometer," 10.1016/j.compag.2014.01.001.

[15] L. Fang et al. "Design of a wireless assisted pedestrian dead reckon-

ing system - the NavMote experience" IEEE Transactions on Instru-

mentation and Measurement, 2005.

[16] J. Kappi, J. Syrjarinne, and J. Saarinen" MEMS-IMU based pedestrian navigator for handheld devices" In ION GPS, 2001.

[17] N. Karlsson et al" The vSLAM algorithm for robust localization and mapping" In ICRA, 2005.

[18] L. Klingbeil and T. Wark" A wireless sensor network for real-time indoor localisation and motion monitoring" In IPSN, 2008.

[19] B. Krach and P. Robertson" Integration of foot-mounted inertial sen-

sors into a bayesian location estimation framework" In WPNC,2008. [20] H. Lepp¨akoski et al" Error analysis of step length estimation in pe-

destrian dead reckoning" In GPS, September 2002.

[21] R. W. Levi and T. Judd" Dead reckoning navigational system using accelerometer to measure foot impacts" US Patent, December 1996..

IJSER © 2015 http://www.ijser.org