V^5 ^tf Lit ttl X A
A System of Detecting Walkers Using Dynamic Bacl^round Subtraction
Abstract
We proposed "Dynamic Background Subtraction"
system for detecting walkers from mobile camera scenes, in order to prevent traffic accidents among automobiles and the walkers. This system separated the walkers from dynamic background of the scene. The system was based on the fact that front street scene of the mobile camera extended from one point: Infinite point. Analyzing the scene extensions, current scene was precisely predicted from the previous one. Stereo-camera provided us depth information for the scene prediction. The difference between the predicted scene and the real one removed background objects in Region of Interested, where the
walkers run out into street in the scenes. In order toreduce the background noises made by vibration and/or curve movement of the automobile, the predicted scene was shifted and subtracted from the real one.
Morphological operation indicated the walkers' residual.
The proposed system was characterized by its simplicity in principle and high potentiality in easily realizing the system with low cost. The method was able to prevent the miss-extraction of background objects and un-extraction of walkers. In this paper, the principle and the procedures of the method were described, and the experimentation which detected walkers from mobile scenes by using the method showed its utility.
fc(i @il]6^(C^ U—dr ^ 7 5 ^ X A75s^^ $ tlT t
tz. u — V b ^ ( d PCS[2]^T'|lfflfb
<fc >5 t, 3;^ h;6^{Sv^fc4t»,
y f y h • -yd V t'i/ 3 A[3]
A $ tiT V^-5.
jE#
Masaki Kasahara
E-mail: masaki.kasahara. 4h@stu.hosei.ac.jp
^ It tti 5yc M17^^ o T V^-5.
^ L^tT#(75fl^G«ai t
y (D X 0 y X,
fe6[8]. Mmy^y(DXo\cy?iy^mm<
ztiizx
•9.
:i Mdm:Scxitz(Dy^y y y y yicx
^ t x , 1 y
mxh'O, ^(D^m'^%^yyyiry:^y\cX-^xn^.
R01(Region of
Interested)!^T', U—Ajij
roB#, M o y - y
-r^.
MMti
=nmxh^ 1 ,^75^ib«(^t't(-J£;55orvK.
7(79ga®^[S]#{cJ;oT
^:b5 /5\ ^COIEE <t V 6 VT'
^ta-rs.
LT, ^ ym{t Ltc. I
7 u-A(|/30 2 ^(Dmm^h,
7l,7D—;^J;oTWf:.^.c07a—h ^.
m'o-<
h>'^^l^< fc46, h/KOX^ $
^ro^ldjiitSrTlHlS:'^^
V u~ f&^y, ^<jykh;^.t^^^.oy{tLU^
I&#i--S. CtiSrn7^-A|ti9jlL, mf|SltO^I^^,-^,^VL.
0 1 e,^.T'^L/c
-nibco,^,{±te
%;t, ^/\^'-y\H<y>±X(D.u t t
Iii>:t7)5*:{l)(7)ct 9j-:^-a:-5.
5 = minZ^iS?«i
ajXii - yf/ + /?/U'+Pj'
a.=yizi n^^'iy-^yi X,-X'^l x,-x
< y —->3 V{cJ;o
rM^^*fli|(cMiE-t^[ll][l2].
^Sy>.r~^/ (*21>>'2l)
V*^br^ (^'5^)
</ju—3^1
(•^ni.ym) \ Nv ..
y;u-:^4,'' ^)-
./•iT
0 I
0 2 ^ 13/vyX2 — tmWM.^^^^m\^fci,oy X(DM\^X(r):t'fy-.<l3jVyu—\t., M['Sis.'^^
t<Dm%^m'm\m7r^'r^fz^y, \
2.2.
oyx% X(D^m^''^¥fi'^xt{^^'0 \ y u—J^fi^\zi6 h ^ y u-j^'xm^^'^-^<r)\iL
®iyo75\ ik<x>y y—1^X13 y y(DW.TMicx-:>XWi{c^W]
•t^t^'D, mi^co.^(r)^mPo^ i3;^yco
im^o, W;,bi3y^y<T)m%^z, 'mr)3
\^\t(OmM^L, WQ^W^(D^W)^^AZt{^, l3/y(D)^
fT:^l^!i^^/oto5!£-^_b,ti-^. mi X'O, AOlPot AOZqWq. J^XlAOlPitAOZiWiit^mXh'O,
\Wm.'^Jt.'^.PoRr^Pi(Dmi^M'^X(D^{2)(DX 0 (d^-lt6.
[fo-k
_f "2'
/p, I
f ~Z-AZ' Wo..-'"
ti/--yCO
0 3
z - Wf:$r ^iJ a fc J6f7)lt^/Pi//Po ^
l3/y (g®)C) (D'^IM^v.)y/d«';(os^rHiPBii5s^4t(=i/3omt-r?it, ^{2)X'ok{t
^i3)xm-rxti)'^x'^^.
! ?_
7^ Z-AZ Z-vAt
^molOH -ST
•5r<$a.^d#4-a)#a3{y^¥W)-^? 'OM'^A
ip)L '9
^P'5#h^^^^flv^^;:4a)(q)^
?:^mm^
'xmm^a)\^
'-r-AC\ vijf(B)i
LEl
Ci^
x:f\-c^^7]mmo)mm'r-^ 'n
'i(^9AnMimm^a>99Z-0 vm-^AA
mm 'nii^mm^mm-'^AA^mmr-^icm
*94-3|y^a-0lf9?^5^l^]ia)^
(9) 'dl
^dl-
^dl
Z
=
•9
^ X':\m<^M'^AA
a)i4i^^f^a)mm94-i^^)m4\
'r)ifY?f(9):;^-^¥
'9$:^
y(/iAX(9)'^a):g
'x^^^^(e)::^?)^viy/sot^ :fo?:2.
^^^9^^
;-5./X5j§
i^a>^—6-^9-^#m::)SaHI
^^
a)
¥»
i*
rz
012 081 OSl 021 06 09 0£
0
'aa)9q:^:i.c.:r-)5^g}Y-x^^a)6K/^:
ligj
':f}/::^xc/'9EI
•4-^rKq)2 EI?)[!0cz)^]l!v^X^
i^HE: 7
-i^m^M: ,,
/M^.(s) p
, -^x-^x
, ^1= '-0j=
2
a)-'x:^^x -91^9,^^55^(5);;^
'?94-ssa-vic.r)z^(t')::^
R-X (1^)
*94i-^':i-::^(t7)a)i^4)-'^?'*::)X<^
'C)
^xrm^\M-'M-'ov
-'X-'J-'OP'^'^
'MVm'OF:
?' C,T
£El
•9X%
^•'x'^'ci>Tmm^^^^wa)mma>T^m^
'n?q^
(^
'z^iggscp4#;\(o?^K
1^c^iJfii^^Y-^'i S
El
•9r#atm50^MvrX;:^-9$^50^
|!}^.^a>f}!il-~/ix:)-^T^)l^ggcp(^^4i^
*Ci Kf^cpo3
miommmm'ct
q>^a)i^¥3^9T^)^gS??if t^EI
91 frl 21 01
q/iuj|09 q/ui5|0t7 q/iu>i0£
gsa)Mv:x^:^9T::i6 '^Piziggj
^ la
•9¥^4.C'Tr:
'CiT(£)::^
(a) -JL (b) v--7'xi!iilft
(c) (--i^lS) (d)
HI?
Roi mmt
W 30x20cm (omu^ 5m t, 640x480 IC^3 i 6 # gg c/5 * c?5 t" ^ -t /L- (7^ ^ ^ ^ t-hsy L r-c.
MS(r>^yyit, ;0 ^ 7 im co!i^7)i
ROI 5m 3|ScOB^|t I5m) I
2.5m 375 b°^-ir;K
1.7m Ci+SS-r-S 255 f^-ir/i^, t
tz. Ix^Lfc ROI 9(w:T^-f-.
150 135 120 A 105
15 20 25 30 35 40 45 50
[US
W^Lfz ROI
2.6. m^t^m
i^V(D\^]SitUcOt^y±J\^(D RGB i$.(Dm.^ t ") , ^(T)^
13:^ L fc r41i^ T0ti(±*11 f: t ^ L l^^-r 6. X -blHl^, 9 m t LX9^6.
2.5 mx'^
^fz ROI -T-?iy4fijcLfc:^f:(C(±, v-'X yii. 0
f)'^h^(Dx, tfz, McomW]
'^^:^-y'\zX6%^fx(r>n-^^ii^hi-tzi^, 2.7 mxm^^
l?±75^ ^.
^ fiSw m ^ ^-c % -r 5 it f: 35 5-> (7) ?g M ifi 75 >
t r a ^ .
mth-t^fzib. ?3e^355>^ei: L/z 2 fi
2 l x . x/i'y
^C7)^m(±.
-f 9 MT-ifyUTi^
-ecofg^
9 6. 7±-
x=^y•:fX(DW^mum-i 1 [Hi. i6
[Uffofc. m
^(7)7
m 10 io(a)i±. m.tE(o
y ]^-j^mutm 7(d)t7)i^'iiiimM®{ti(o^5>T'5£fig$n
fz. m^mitxh^.Ayt'm^,
wicw^yo-z^y^izx^^^mo ttlldi'efc'9, I0(b)'C"/j^^ < ®tfj[±*oT^^/c^ff#c75^^
1 -^(Di^^mM^izm^^tz. ^mz I0(c)-t:'coa^(7)
iilU^. 10(d)(75 J; -) iz^y I/—(a) mmmm (b) x~x:^yym
(c) (d)
[gjio
2.7.
^^•t. ZCDi^yyit, 1 #(30 7U-A)rBl(::4o{t-5^H'=-^
(DM.^ h \yy> t ,
•5tgtl]'l'ico^:l^l±_tT:^f^T'±5 t'^-tr/L', ^;&;^|p]-r-
±101' ^ ir/u- i: o fc. r t O l S f f l ^
i"5. ROI
r« = p
'^5feco|fL©l:
1 _ - :: -I ::,
550 560 570 580 590 600 610 620
Hill
HI 12 It, HI \o{a)(DBmmi^X'X)mi^W]W:{Eoy^m^
HI 12(a);t¥fT#®]ML-eM5>^tto/i
'^mx^'O, 12(b)-eit
X"^X^^^(D-f}'^t>^^^. ?iM'^l±(a)7!i^ 17.2%, (b)7)'^ 3.5%
Xf^-otz.
(a) WiEtiL
HI 12 ¥'^J^
(b) ^lE^Ki
mm<D¥n^wn^x?>mmnmit,
tm&.x^, mm
<D\im.-t^t LT^T'tS.
Hi 13(a), (c), (d){t±'lS--cOML#7i^i7, 13(b), (e).
nx^tzx^xxtz^, ROI
^n^tz. ^ 1 \z^'y-yX(Dmimh(r^''^m\z.X7^B HI 13 i:* 1 t'), t i o - x (D 2 -y-yxij, ¥ff^W]^'o
^Xztx^x^fz. R0l^m^X6Zti^X^X,
± 5:: <b 6.
(a) ±"9 (b)
(c) (a)cOME4^ (d) (a)(D^lE^V'^¥
(e) (b)<75MlE7'^ (f) (b)(7)^Xib'l^m
HI13 ^~ytts--(D^LmxvmzX6Bmmm
^ 1 ^^y—y(DB^mm
'y-y WiEf£ um¥{%) MiEfc ^) Bm¥(%)
±V)m 7.12% 2.58%
±7—y I 18.27% I 5.86%
3.
3.1.
XXtzi^), y y Uyt:^ / y t T l ' ^
^^mx(D\^mx,
m^n^tz. %m.xi.txK>'^<(r)mi:^^\k\Ex^fzi^, : t j / y y y X ^ ^ 3 \^X—J^^iiA^fzW:m^\^^Efz.
B;tm-Cw^;coB#igit4sJ;'e io.8km/h,
yyXf] / y \ t FUJlFILM^tCO FINE PIX REAL 3D,
7 U—Amt 120 CPU it Intel Core 15-
2520M CPU(2.50GHz)T'S5§. 65
3.2.
m 14 l4{a)(±-cO||gi(Cio(t5®l[!S
ji.-SiROlT', Z(D KO\^X'<DWil^^^\^Lfz. 14(b)
14(c)li:^^f:SP^(C;^^ < LfzmX'h V, &(D^^
I4(d)li mf^^f^tax^^t^^^-Dfzmxho.Ax^^Ltzmr^
[1 I4(d)(7)j;^;::, (4®
\20 l^kxh'o, i:
5.8%t^£^tz. ROI l^-e, CO 2 11(75^
ff# (m 14(a);fe-)
il7)^iltiaf5i:'®{tco^^T#(0^®j*/5t/>7'^< ^
^ffiJ;''), m \4{d)X(DXi^W>(oWn^itti^ yXhist.
X eSrnmtltzmm^mmLXts'O, i:^yt(Dmm esm iift^^^^ttlcO[5I)t-efca. CcOgg^Ki. 80km/h X^.
X l .
tH t 9 - 5 — 6 0 k m / h i^TT'CO^ttHli^
k^-^mxh^. Z(DztXh.
ff # It ttj (OW CO I ^:: /)^TP $ tL f;.
(a) ROI (b)
(c) n^mm (d)
4.
i^yyM.bLx,
XLfz. X(Di^yTM.Xit, /y(D'^^X>h6 1
:Lt^^\\J^LX, 1 ARijco®
itxh, / y
XXmXf^Ltz. XLX,
u—A®{t^M5>U, • i/—
L, %m.Xit\^b/^.}zX>
o/ia6,
4i-. ^rp1^®3^0#coM5>#&<o?iS3i:t^lfec7)ai
X fife
[1] "3:2ia^^&^<-T^fiic24ii^it)is>", m\m
W>\UW>1^, alSF-fllSfJ. pp.18, 2013.
[2] S. Tokoro, K. Moriizumi, T. Kawasaki, T. Nagao, K. Abe and K. Fujita, "Sensor Fusion System for Pre-crash Safety System", IEEE Intelligent Vehicles Symposium, pp.945- 950, 2004.
[3] T. Tsuji, H. Hattorl, M. Watanabe and N. Nagaoka,
"Development of Night-Vision System", IEEE Transactions on Intelligent Transportation Systems, vol.3, No.3, pp.203-209,2002.
[4] "x/c/w- y\ a
pp.12-15. 2010.
[5] N. Dalai and B. Triggs, "Histograms of oriented gradients for human detection", IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pp.886-893,
2005.
[6] M. Oren, C. Papageorgiou, P. Sinha, E. Osuna, and T.
Poggio, "Pedestrian detection using wavelet templates", IEEE Computer Society Conference on Computer Vision and Pattetn Recognition, pp. 193-199, 1997.
[7] A. Broggi, M. Bertozzi, A. Fascioli, and M. Sechi,
"Shape-based pedestrian detection", Proc. IEEE Intelligent Vehicles Symposium, pp.215-220, Dearbon, USA, 2000.
[8] K. Toyama, J. Krumm, B. Brumitl and B. Meyers,
"Wallflower: Principles and practice of background
maintenance", Proceedings of the 7''' IEEE International
Conference on Computer Vision, pp.255-26l, 1999.
[9] C. Harris, M. Stephenes, "A combined comer and edge detector", Proceedings of the 4th Alvey Vision Conference, pp.l47-15l, 1988.
[10] B. D. Lucas and T. Kanade, "An iterative image registration technique with an application to stereo vision", Proceedings of the J98I DARPA Imaging Understanding Workshop, pp. 121-130, 1981.
[11] J.g. Frayer and D.C. Brown, "Lens distortion for close- range photogrammetry", Photogrammetric Engineering and Remote Sensing, vol.52, pp.51-58, 1986
[12] Z.Zhang, "A flexible new technique for camera calibration", IEEE transactions on Pattern Analysis and machine Intelligence, vol.22, pp.1330-1334, 2000