Vol.37, No.2 ( pp ) 2) A Study of a People Tracking Method Using Invisible Visual Markers in Indoor Environments Susumu SHIRAYAMA and Rie

Size: px

Start display at page:

Download "Vol.37, No.2 ( pp ) 2) A Study of a People Tracking Method Using Invisible Visual Markers in Indoor Environments Susumu SHIRAYAMA and Rie"

次冯
4 years ago
Views:

1 Vol.37, No.2 ( pp ) 2) A Study of a People Tracking Method Using Invisible Visual Markers in Indoor Environments Susumu SHIRAYAMA and Rie OSAWA ABSTRACT In many worksites, there are some tasks that each worker s location will affect the work efficiency, the safety of work, the frequency of mistakes, the physical and mental workload, etc. The relationships between the location of people and the efficiency of the work, etc. have been investigated using personal trajectory data. The data is obtained by people tracking methods. Although the tracking methods have been developed by many researchers over the decades, several issues remain. One of the biggest issues is the lack of capability of individual recognition. Another of the biggest issues is to protect the privacy of the people involved in the tracking. In this paper, first, the existing methods of people tracking are surveyed and the necessary requirements for tracking the people are shown. And then, a new people tracking method using invisible visual markers made of a near infrared LED is proposed. We attempt to recognize multiple people by our method. It is shown that two peoples can be successfully tracked separately in our laboratory, and the privacy of the person is protected. Keywords : People tracking method, Invisible visual marker, Individual recognition 1),2) * ) , E- mail : sirayama@sys.t.u-tokyo.ac.jp) 2) , E- mail : rie-u@nakl.t.u-tokyo.ac.jp) 1 LED 3)

2 4) 5) 2.5m 2.5m 2.5m 0.8m 0.5m/s1.8km/h Miyazaki Kasama 2m 6) ) RFID Sorensen RFID 8) RFID RFID 9) RFID 10) RFID RFID 11) GPS LAN 12) 13) GPS GPS GPS GPS Nirjon GPS 14) 15), 16) Liu 17) 18) Yang 19) 19 RFID 20) PDRPedestrian Dead-Reckoning PDR Yoshimi 21) PDR RGB-D 22) LED 23) (a) (b) (c) (d) (e) (f) (g) (h) (i) (j) (k) (l) (m) (n) 2

3 (a)(n) 3) 7 RFID RFID RFID RFID 13) x 20 PDR PDR LED 23) 24) 25),26) LED 27) 24) 25),26) 23),24) 380nm760nm 3

4 ᒣ 㸪 ᜨ ࡉ 㸬 ᪉㸪 Ἴ㛗ࡀ 400nm㹼1100nm 㞴ࡋࡉ ၥ㢟ゎỴࡉ 㸬ࡇ ศᯒ ౽ࡉ ᅖ ᙳࡀ 㸬Fig.1 ࡍᖏᇦ 㐪㸪ᥦ ᡭἲ ࡁ Ⅼ 㸬 ࡋ㸪㛫ㄆ ࡁ ࡀ㸪 ᙳ 㻟㻚㻞㻌㻌䠎䛴䛾 ᡭἲ㻌㻌 ᮏ 㸪Fig.3 ࡍ IR IR ࢱ ࡁ ග స ฟࡍࡇ ࡀ ࡁ 㸬 ࡋࡓ㸰ࡘ ᡭἲ ᥦ ࡍ 㸬どග 㛫どග䠖䠄380䡚760nm䠅 እ ᅗ Mi IR 㸪Cj ࡍ㸬どග እ IR䝬䞊䜹䜹䝯䝷䠖䠄400䡚1100nm䠅䜹䝯䝷 M1 M4 እ M7 M5 M2 M3 M8 M6 M10 M11 M12 M9 እ LED C1 Fig.1 The difference of the frequency band that is C2 ᑠᆺ䝡䝕䜸䜹䝯䝷 visible to the human eye and the camera M15 M13 M14 ᮏ 940nm እ ᇦ LED ࡋ㸪 እ LED ᵓᡂࡉ సᡂࡍ 㸬ࡇ ど㸬ᮏ 㸪 እ ࡇ 䝡䝕䜸䜹䝯䝷㸪IR㸦near-Infrared Red㸧 ࡍ 㸬 C1 ど ࡍ ࡇ 㸪ప く ศᯒ ィ ࡀ 㸬ࡋ ࡋ ࡀ 㸪ど ௨ እ ሗࡀ ࡓ 㸪 ࢩ ಖㆤ㸪 ሗ₃ὤ ᅇ M1 M2 㑊せ ࡓࡍࡇ ࡁ 㸬 IR䝬䞊䜹䝬䞊䜹 ࡑࡇ 㸪 እ ࢱ ࡍ 㸬 እ ࢱ 㸪ᆏ ᱵ 23)㸪ᅵ᪉ 25) ᥦ ࡉ 㸬ࡓࡔࡋ㸪ࡇ Fig.3 Schema of method (A) (upper) and (B) (lower) 㸪㠀くᛶ㸪 ࢩ ಖㆤ㸪 ሗ₃ὤ ᅇ㑊 ࡋࡓ 㸬 ௨㝆㸪 እ ࢱ IR ࢱ ࡍ㸬ᮏ Fig.3 ᅗ ࡍᡭἲ(A) 㸪ᐊෆ ቃෆ IR ࡋࡓ ࢱ 㸪900nm ௨ Ἴ㛗 ග 㓄 ࡋ㸪 ࡋࡓ IR ࢱ ࡁᑠᆺ ࡀ 㐣ࡍ 㸬ࡇ 㸪 ᙳࡉ ග ᙳࡍ ᪉ἲ 㸬 Ἴ㛗ࡀ 900nm㹼1100nm ᅖ 㸬Fig.2 900nm ࡓ㸪Fig.3 ᅗ ࡍᡭἲ(B) 㸪 IR IR 㸪 ࢱ ᙳࡋࡓ ࡋ㸪ᐊෆ ቃෆ タ ࡋࡓ IR ࢱ ࡁ ࡍ㸬 ᙳࡍ ᪉ἲ 㸬 ᡭἲ(B) ᚑ᮶ ศᯒ ἲ ᘏ㛗㸬ᡭἲ(A) ᥦ ࡍ 㸪ᐇ㝿 ᕤሙෆ ᐇ ࡋࡓணഛᐇ㦂 ᯝ㸪ᐊෆ ቃ タ ࡁ ሙ ࡀከࡃ ࡇ ࡀ ࡋࡓࡓ 㸬㻟㻚㻟㻌㻌㻵㻾䝬䞊䜹䛾㻌㻌 ᆏ 23) 㸪ᖖ Ⅼⅉ Ⅼ 㸰㢮 Ⅼⅉ ࢱ ᣢࡘ 1 ಶ IR ࡋ 㸬IR Fig.2 Without (left) and with (right) a IR filter RS232C ࢱ ࢫ ࡋ㸪ไᚚ PC ಙ Fig.2 ᅗ ᙳࡉ IR ࡅ ࡗ 㸬 ᐊෆ 㸬ࡇ ࡓ 㸪 ሗ₃ὤ ၥ㢟ゎ IR ᖖ Ⅼⅉ ࡋ㸪᪂ࡓ ධᐊࡋࡓ Ỵࡍ 㸬 ࡓ㸪 ᙳᑐ ᅖ 㛫ࡀ ᛶࡀ IR Ⅼ ࡍ ไᚚࡍ 㸬Ⅼ ࡋࡓ IR ሙ 㸪 ᚰ ᢸࡀ ῶ ࡁ 㸬 ᐃ 㛫ᚋ ᖖ Ⅼⅉ ࡋ 㸬 እ ࡉ 㸪 ฎ ༢ ࡉ 㸪㛗㛫 ᑐࡍ ศᯒ 㛫 ഐ ᐃ 㸪 ࡋ㸪 4

IR PC (B) IR LED IR IR IR LED PIC Peripheral Interface Controller 12F675 Arduino

1,,p i,, p N, } p i 0 1 N p 1 p N N IR p M {p 1,,p i,, p M, }p i 0 1 M p M p N p

5 IR PC (B) IR LED IR IR IR LED PIC Peripheral Interface Controller 12F675 Arduino PIC LED Fig.4 IR LED IR LED 0 RGB YUV Y LED Y Y 255 Y Fig.5 LED IR IR Fig.4A prototype of IR marker (upper) and a circuit of the flasher (lower) IR p N {p 1,,p i,, p N, } p i 0 1 N p 1 p N N IR p M {p 1,,p i,, p M, }p i 0 1 M p M p N p M p N R pp (i) R pp' ( i) p p' i i j (1) j M p M p N IR Fig.5 An example of the image processing for recognition of IR marker and labelling the marker Fig.6 5

6 t f t E t f t E Fig.6 Schema of frame rate and the exposure time in a video camera LED t f PC t f Fig.7 A p M p N p M A p N Fig.7 False detection of LED flashing p 12 {0,0,1,1,1,0,0,1,0,0,1,1, } 1 LED 12 p 12 Fig (a) Normal pattern (b) Example of image frames detected falsely Fig.8 A part of the video image for the pattern p12 t t Fig.8(a)(b) p M {0,0,1,1,1,0,0,1,0,0,1,1,0,0,1,1,1, } p M {0,1,1,1,1,0,1,1,0,1,1,1,0,1,1,1,1, } PC p N p N C N p M (1) p N C N M p M p N C N IR p 12 p 12 C 12 C 12 C 12 {0,1,1,1,1,0,1,1,0,1,1,1, } p M p M p 12 C 12 (1) M p M p 12 p M C 12 Fig.9 M 12 M=12 p M p 12 {1,1,0,1,1,1,0,1,1,1,1,0, } C 12 C 12 p 12 p 12 IR Fig.9 Time variation of the correlation between p M and C 12 LED 1 LED 6

7 (A) 1 LED IR 2 LED 10m 2 IR (A) IR IR IR IR Fig.10 Table 1 IR 0.75m 5Hz 3/30 3/30 p 6 {1,1,1,0,0,0, }2.5Hz 6/30 6/ Hz 9/30 9/30 IR LED 0.75m 5) 0.8m Table 1 Types of IR markers () () 1 1 5Hz Hz1.67Hz 3 5Hz2.5Hz1.67Hz Fig.10 An experiment using the method (A) 1.1 m/s4 km/hfig.10 1m IR IR IR (A) Fig.11 IR (A ) IR IR (A)(A ) IR M6 M5 P2 C1 IR P3 M3 Fig.11 Schema of method (A ) (A ) IR IR Fig.12 Fig.12 Fig.10 Fig.12 LED Fig.12 Time variation of the number of lighted LED, and the IR landmark estimated by the method (A) M7 P4 M8 M1 C2 P1P P5 M4 M2 B C D A 7

Fig.12 IR IR IR D ABCD IR IR (A ) (B)Fig.13 Fig.13 IR IR Fig.

15 Example of the estimated location of a person at a time IR IR LED IR Fig.

13 Glass-type video A 0 30 60 90 120 150 180 210 Fig.

8 Fig.12 IR IR IR D ABCD IR IR (A ) (B)Fig.13 Fig.13 IR IR Fig.14 IR 1 LED A Fig.12 P3 IR P3 Fig.15 IR P1 P2 P3 P4 P5 t Fig.15 Example of the estimated location of a person at a time IR IR LED IR Fig.16 2 m IR Fig.17 Fig.17 Fig.18 Fig.17 Fig.18 IR LED Fig.13 Glass-type video A Fig.16 Helmet with IR maker Fig.14 Time variation of the number of lighted LED (A) (A) (A ) IR (A )Fig.14 Fig.17 Location of the IR marker 8

9 S 1 F F 3 Fig.18 Trajectory of the IR marker F2 F 1 S E2 2 G Fig m 4.5m E 1 O I Fig.20 Trajectories of two peoples IR S1 O I E1 IR Fig.19 Environment of the experiment 1 1 p 6 {1,1,1,1,0,0} 4/30 2/30 LED IR 1 LED IR IR IR Fig.20 IR G IR IR IR F F1F2 F3 3.4 IR F1 F2F2 F3 IR IR IR 28) LED LED IR (A)(A ) (B) LED LED (A )(B) (A ) 9

10 (B) (A ) (B) (A )(B) 1) 2008-MPS-72 (6), (2008), pp ),,,, 2009-MPS-73 (50), (2009), pp ) (2003), pp ) DEWS2008, (2008),D ) No.2004, Vol.112, (2004), pp ) Miyazaki, T. and Kasama, Y. : Multiple Human Tracking Using Binary Infrared Sensors, Sensors, Vol.15, (2015), pp ) Vol.57, No.10, (2016), pp ) Sorenson, H.: The Science of Shopping, Marketing Research Vol.15, (2003), pp ) 56 2(2008), pp ) 21 (2007), 2C ) RFID Vol.73, No.630, (2008), pp ) indoor.locky: LAN Vol.2010, No.4, (2010), ) Vol.57, No.5, (2016), pp ) Nirjon, S., Liu,J., DeJean, G., Priyantha,B., and Jin, Y.: COIN-GPS: indoor localization from direct GPS receiving, Proceedings of MobiSys 14, (2014), pp ),, 12, (2011), pp ), 13, (2011), pp ) Liu, J., Hirota, J., Wu, C.-X., and Aoyama, K.: Monitoring System for Advanced Construction Management (Integration Methods of Monitoring Data to Generate Virtual Job Shop), Proceedings of ICCAS, Vol.2, (2013). 18), Vol.37, (2009), pp ) Yang, C.-J., Chou, T., Chang, F.-A., Suu-Yuan, C. and Guo, J.-I.: A Smart Surveillance System with Multiple People Detection, Tracking, and Behavior Analysis, Proceedings of 2016 International Symposium on VLSI Design, Automation and Test, (2016), pp ) 2009-UBI-24 (25), (2009), pp ) Yoshimi, S., Kanagu, K., Mochizuki, M., Murao, K. and Nishio, N.; PDR Trajectory Estimation using Pedestrian- Space Constraint: Real World Evaluations, Proceedings of UBICOMP/ISWC 15 Adjunct, (2015), pp ) RGB-D PDR PRMU (2016). 23) ALTAIR: IR EVol.123, No.8, (2003), pp ) TVRSJ, Vol.10 No.3, (2005), pp ) LED '05 (2005), 1A1-S ) Sohn, B.Lee, J., Chae,H. and Yu,W. : Localization System for Mobile Robot Using Wireless Communication with IR Landmark, Proceedings of the 1st International conference on Robot communication and coordination, (2007), pp ) ID LED Vol.51, No.1, (2015), pp ) Vol.36, No.143, (2016), pp

<4D F736F F D BD90AC E92C091DD957393AE8E598C6F89638AC7979D8E6D2E646F63>

<4D F736F F D BD90AC E92C091DD957393AE8E598C6F89638AC7979D8E6D2E646F63> 5,000 20 18 20 10 12 60 60 ၥ ࠚ 㛵ࡍ ḟ グ ࡕࠊ 㐺ษ ࠋ 㸯 ࠊ ཤ ቨ ࢡ ࢫ ᖺ ཬ ᖖᦆ ศ ᙇ ࡘ ࠊ ࡉ グ ࢢ ᚑ ᙇ ᢸࡋ ࡅ ࠋ 㸰㐣ᖺ ࡓタഛ ࡗ ࠊ ࡋ ㈤ ఫᏯ タഛ ࡋ ሙ ࡀ ࠊࡇ ሙ ࡀᨾព 㐣ኻ タഛ ᦆࡋࠊ ࡋ ࡋ ࡗࡓሙ ࠊᚑ᮶ᶵ ࡋ ࡓ ᅇ ࡉࡏ ࡓ ࡀ ᢸࡍ ࡁ ࡁࡀ ࠋ 㸱 ᢸ ㄢࡍ ࡘ ࠊࡑ ࡍ ᚲせᛶࡀ