о뷢չ
JOURNAL OF COMPUTER RESEARCH AND DEVELOPMENT
1999Vol.36No.6P.690-694



Ӿͳһʾ
⸣ں鴨ԬΤ
ժҪͨõĻʾʽΪF=K-TtݡRK-1УKΪڲR,tֱΪתƽʸ.ת˶ӦĻ󲢲ڴ˱ʽ֮.иһµĻʽF=eݡKRK-1eǼ.ʽͳһ˴ת.Ҫĩ궨£ɴ˱ʽʹڲ˶õ.
ؼʣӾ⼫Σ
ţTP18
AN UNIFORM EXPRESSION OF THE FUNDAMENTAL MATRIX
WU Fu-ChaoYU Hong-ChuanYUAN BoWEI Sui
(Institute of Artificial Intelligence of Anhui University, Hefei 230039)
AbstractThe fundamental matrix's expression is usually F=K-TtݡRK-1, where K is the intrinsic parameters matrix of camera,and R and t is the rotation matrix and the translation vector of camera respectively. The expression is not valid when using the pure rotation. In the paper here, a new expression of the fundamental matrix ,F=eݡKRK-1, is presented, where e is epipolar point.Net only the expression is valid when using the pure rotation, but also it predigestes the following problems, such as, the intrinsic parameters matrix, the motion parameters, and structure recovery from uncalibrated stereo rig.
Key wordsactive vision, epipolar geometry, fundamental matrix
1
ӾУڲߵر궨ǲܼٶڲ֪ģҲ˵Ӿȡͼͨδ궨.δ궨ͼƥϵڲ˶Լع⣬ʮҪĽɫ16. ͨʹõĻʾʽΪF=K-TtݡRK-1УKΪڲRtֱΪתƽʸ.˱ʽȱ㣺 ת˶ӦĻ󲢲ڴ˱ʽ֮У . ĸһµĻʽF=eݡKRK-1eǼ.ʽͳһ˴ת.ǵоδ궨£ɴ˱ʽʹڲ˶õ.
2͸ģ
ڱǿǵģΪСģ.ڴģУXC͸ӵƽϵصm. ԹҴֱƽһߣƽ潻c.ƽϵΪc-xyӦϵΪC-XYZZغϣX,Yֱƽx,yᣬϵC-XYZغ.ڲΪ

򳡾X=XYZTͶӰƽϵm=xyT㣺

 m=uvwTu/w=xv/w=yУ
m=KX1
˶תΪRƽΪtʱµϵΪC-XYZ,C-XYZ֮ĹϵΪX=RX+t򳡾X=XYZTڵǰƽϵm=
(u,v,w)\+TΪ
m=KRX+Kt(2)
3⼫
ڿӵ¹۲ͬһͼ1ʾ.CCֱΪ12ĹλãCڵ1ƽIϵͶӰΪeCڵ2ƽIϵͶӰΪe䣬ǳΪ⼫.ƽI(I)ͨ ee䣩ֱ߳Ϊ⼫.ӾУ⼫ԼƽIһmƽIϵƥmλ⼫lmϣƵأIƽһm䣬ƽIϵƥmλ⼫lm.lmlmΪӦ⼫.


ͼ1⼫
ӰռڣƽϵֱlӰʾ.mm⼫lmlmӰԼΪlmlm䣬lm֮mһԱ任7ݣ
lm=Fm
F2ľ󣬳Ϊ.
mƥm⼫lmϣ
mTFm=0(3)
3ʽתã
mTFTmT=0
mӦ⼫߿FTmʾ.
ʣ
1 FΪҽF㣨3ʽ rankF=2
2 eFe=0eFTe䣽0
3 F һ㳣Ψһ.
4ڲʾ֮Ĺϵ
X=XYZΪһ㣬mI,mIһƥ㣬ɣ12ʽ
m=KX,m=KRX+Kt

m=KRK1m+Kt

K1m=RK1m+t


tݡt=0
tݡK1m=tݡRK1m

mTKTtݡK1m=0

mTKTtݡRK1m=0
rankKTtݡRK1=2
FKTtݡRK1(4)
ǻ[t]RǱʾ.
t=0ʱtݡ=0F=0ǻ.Ϊת˶ʱ󲢲ͳһڱʽ4֮.
5ͳһʾ
ڸͳһʽ.
1. 
F=eݡKRK-1(5)
֤.t0ʱɣ4ʽ
F=(KTtݡK1)KRK1
ֻ֤һӵ
KTtݡK1eݡ
ĳʼϵ˶ϵ֮ĹϵΪX䣽RXt˳ʼλX=0˶ϵеΪX=te=Kt

Ѽ㣺




ˣһ㳣ӵ
KTtݡK1eݡ

FeݡKRK1
t=0ʱX=XYZTΪһ㣬m=(u,v,w)TI,m=(u,v,w)Iһƥ㣬
m=KX, m=KRX

m=KRK1m
KRK1=(mij)33(u,v,w)+TŷΪ

ʽеĵ3֪
s=s(m31x+m32y+m33)

(m31x+m32y+m33)(x,y,1)T=(mij)33(x,y,1)T

(6)
ӣ6ʽȥm31x+m32y+m33
x(m21x+m22y+m23)y(m11x+m12y+m13)

-m21xx-m22yx+m11xy+m12yy-m23x+m13y=0
дɾʽ




mTFm=0
rank(KRK1)=3rankF=2Fǻ.
e=(001)T

ݻĻ(2), ֪eǼ.

ʶ.
6ᡡ
ڱУǸ˻ͳһʾֱʾǷǳõ.ǵоʹñĵĻʾʽʹδ궨ͼڲ˶õΪ򻯣Щķ.
ĿõȻѧ͹ҡƻ.
߼飺⸣У195710ڣĿǰҪ˹ܡӾӦѧȷĿѧ.
ں鴨У19696ʿоҪ¼Ӿʵо.Τ룬Ů19464ڣʿʦУ.ĿǰҪӡ¼Ӿģʽʶʵȷо.
ߵλ⸣մѧ˹оϷʡ230039
ں鴨մѧ˹оϷʡ230039
Ԭմѧ˹оϷʡ230039
Τ루մѧ˹оϷʡ230039
οף
[1]Hartley R I. Estimation of relative camera position for uncalibrated cameras. In: Proc European Conf on Computer Vision. Santa-Margerita, Italy, 1992. 579587
[2]Deiche R, Zhang Z, Luong Q T et al. Robust recovery of the epipolar geometry for an uncalibrated stereo rig. In: Proc European Conf on Computer Vision. Stockholm, Sweden, 1992. 567576
[3]Maybank S J, Faugeras O D. A theory of self-calibration of a moving camera. The International Journal of Computer Vision, 1992, 8(2): 123151
[4]Robert L, Faugeras O D. Relative 3D positioning and 3D convex hull computation from a weak calibrated stereo pair. In: Proc Int'l Conf on Computer Vision. Berlin, Germany, 1993. 540543
[5]Zhang Z, Deriche R, Faugeras O D et al. A Robust technique for matching two uncalibrated images through the recovery of the unknown epipolar geometry. Artificial Intelligence. 1995, 78(1-2): 87119
[6]⸣ں鴨Ԭ. Ա궨-㷨. Զѧ1999, 25(6)
(Wu Fuchao,Yu Hongchuan, Yuan Bo et al. Camera self-calibration-theory and algorithms. ACTA Automatica Sinica(in Chinese), 1999, 25(6))
[7]Luong Q T, Faugeras O D.The fundamental matrix: Theroy, algorithms, and stability analysis. The International Journal of Computer Vision, 1996, 17(1): 4375
ոڣ1998-08-10
޸ڣ1998-11-30
