ѧ
JOURNAL OF SOFTWARE
1999 2 No.2 1999



ӾлģӰ㷨о
ľӡڡƽοҡš
ժҪʵʵӦ,ƶŸӵĻ,·гӰˮȻ.ЩصĸӾ㷨³.һµƶ˵Ӱ㷨.ʹģڵͷֱͼ϶Ӱʶ,Ŵ㷨ṹŻ,ԭʼͼӰķ͹.ͬʱø㷨THMRIIIƶϽеʵ.
ؼʡӰ,Ӿ,ģ,Ŵ㷨.
ͼšTP18
Shadow Eliminating Algorithm with Fuzzy Neural Network in Vision Based Navigation
GUO Mu-he YANG Lei TAO Xi-ping HE Ke-zhong ZHANG Bo
AbstractIn past decades, there are many vision algorithms. In practice, the environment in which robot stayed is filled with stochastic noise such as shadow, puddle, and fallen leaves. Those factors damage the robustness of the past vision navigation algorithm. A new method to eliminate shadow during mobile robot navigation is presented in this paper. The presented method gives a detailed description on how to use fuzzy network to recognize and eliminate the shadow in low-resolution image during robot navigation, and optimize the structure of fuzzy neural network by genetic algorithm. Also, the experiment results obtained on THMRIII Mobile robot test-bed are given in this paper.
Key wordsShadow eliminate, vision navigation, fuzzy network, genetic algorithm.
ڹȥ10,ӾĽչǳ.ӾⱾĸɵ,ǺҵһֶԻķʽӾ;һԭܵͳļֶε.ôͳʶҪȷԵֵ㹤,΢,ڸӻȷϵͳоͺҵһȫֺԷϵͳĽṹз.10,ΪҪ㷨Ӧúܶ,Ҳ,,Ȩֵ.֮,ģһнϴ,ͬʱ,ģṩһ,ܹģеĽǿеѧϰӦݴԺͲ.
ģģĿƺ;ϵͳѱΪ㷺Ӧڸ֮У1,2.ϵͳеģͨרҵ,ʵ֪ʶ.,ЩѧֱԶƵ,֪ʶ.ף3ĻģӦϵͳ,ַܹԶʶģ߼,ϵͳȷ.,Щѧģ㷽ʽѧϰЧʣ4.ģ,һӵϵͳҪģԹ,ӶģĸԽϴ.޵ʱڽЧȡһǳҪ.ף5ʹԷķģ.
ӾڻӾռҪĵλ.,㷨³ԳڻӰ,,ӾеҪ.ȵӰˮȵ.ڱ,õͷֱͼ,ʹģķл.
1㷨
֪ʶֱȡģһѵĹ,漰¼Ҫأ(1) ı﷽ʽ(2) ģƵĹ,ụ̃́(3) ЧĿ,еȺıģжоԵӰ.,ȶȺı.ӰˮͼûнṹԵĴ߶,ûʲô֪ʶ.ǵľ߹̻,ӰˮͼɫԹ̶.Ȼ,κӰɫֳһۿǰƺɫ,еɫسɷü򵥵Ĺʽ.ǲѧϰķ,йӰˮɫسɷֵ֪ʶȨֲֵʽ¼,ӶﵽʶĿ.ʶӰģBPͼ1ʾ.,R,G,BֱΪͼأi,jĺ졢̡ɫǿȣҶȣ2ģ,ͼ2ֱͨ,Ϊ
{ai,0ai255aiai+1,i=1,2,...,12},(1)
˸ò㼰һĽṹͨȡͬĲ3Ϊģ,ȡʽ:
If (=p And =q And =r) Then RuleNo=i; i=0,1,...,127
ΪӰбĶֵ,


ͼ1R,G,B·Ҷģͼ2Ӱʶ
1.1ʹŴ㷨ṹŻ
Ŵ㷨ΪHolland,Ӧõ6.Ŵ㷨Ϊ㷺ӦŻ.Ż,ѽֵ󵼵ķʽѡŻ,ʹ㷨Ե÷,,ЩѧӦŴ㷨ĽṹŻѧϰѰֲС.ǵķܵ۵.ṹŻѧϰ,ҪԲ(1)ѡ,Ŵ㷨Ż.
ȾɫSi=(xi,yi,zi),xi,yi,ziΪR,G,B·ģȾɫ{xi,yi,zi},ô,ŴӽйȡĹ£
㷨1. ʹŴ㷨йȡ
1ͼȷȾɫʼ⣻
2ѡŴ,µȾɫأ
(1) CROSSOVERȾɫ1ͦ2ֱͬһλп,ν,õµȾɫ.
(2) ͻMUTATIONȾɫ1ĳ򴦷ı,Ӱ.
(3) λSHIFTȾɫ1ĳ򴦷仯С,ͬһƽյ,С.
3ѡȺ,ﵽֹͣ˳,2.ŴѡŵʶȺFitnessΪ
,(2)
ЦΪЧ,ΪЧ.,ԭʼͼѱӰɼ,нṹѧϰ,տɵõŻģ.
1.2 ѵѡ
ȨֵѵĹ̲ͨBPѵʽ,ҪѡȡӦĴ.,Ȼջĸ,ʹĵԺж,ѧϰ뱣ƫ.ͬʱ,BPĽṹĸӳ̶ֱ.,ڽṹŻǰ,Ӱʶѵĺû,Ҫڸֹջ,ɼо,Ȼѧϰ.ĽǵõȨֲֵ,ÿһȨֵӦһಢһķ.繤ʱ,жͼ֪һ,ȻӦһȨֵ.Ҳ֪ƥʱ,˵绹Ӧµһಢѧϰ.,ʵӦ˵,һʶγӦһʽĹ.
ѡһõľ㷨Ҫ.㷨Ļ£
(1) ӰʶͼȺɫָϵ:ʵɸѡ,ȡ¼ͼ
(1.1) ȣѡ׼ȷ
Y=0.22R+0.587G+0.114B,(3)
һͼƽȡΪ
(4)
(1.2) ɫͶȣȡ졢̡ɫԱͶΪ
(5)
һͼɫƽͶȷֱΪ
(6)
,{MY,SY,MR,MG,MB}Ϊָ꼯.
(2) ģƾȡP={p1,p2,...,pm}Ϊͼ,ģϵF(UU):
(2.1) Է (u,u)=1,uU
(2.2) Գ (u,v)=(u,v),u,ԡU,
Ϊģƹϵ,R=(rij)55Ϊָ꼯ģƾ,ԽԪΪ 1,rijʾͼpipjѡָ꼯Ƶĳ̶ȣϵ.ѡָ꼯Сrij,
(7)
pikʾiͼĵkֵָ.
(3) ࣺģƹϵF(UU)Ϊģȼ۹ϵ*F(UU),
(8)
Ԧˡʣ0,1ݽȡR*R*,ջͼ񼯵ķ.
2ʵ
ڲͬڣġͬ硢硢磩ȡ100ͼ(ͼ3е(1)ʾ),1.2о,ȡ=0.5,õ5,ͼ3(a)(e)ʾ,(e)Ϊˮ.ѧϰ5Ȩֵ,ͼ3е(2)(3)ʾΪòͼõӰˮʶ,ͨӰͼԭʼѹͼԵͼ,ͶӰϵȶ֪ʶ,յõĵ·ϰԵͼ.ʵл,õͷֱ64*64ѧϰ,ڸ߷ֱ256*256¹,ʶʼ.64*64ֱ,ʶʱΪ250s(PC-586,Ƶ133Hz),ʵʱԺܺ.


(a) :1%(b) :4%(c) :2%(d) :2%(e) :1%
е(1)(3)зֱΪԭʼͼӰʶͶӰ
ͼ3THMRIIIƶʵƽ̨ϵ5ʵʵ
3  
ڻӾϢʽƶ˵о,ĿǰҪ㷨ĸʵʱԵì.ӵʵҪӾϵͳ׼ȷȡĵ··꣩,ƱҪڼѰ,֮ͼеϢģ,ʹòôͳĻӾͼͻ㷨ԵԼ.һ,˶ŵĻǲ֪ϵͳнϴӳٵ.ͨоʵ,巽ʶǽṹȷȣǰ;ֶ,۴оʵüϿ,.,Ŵ㷨ṹŻһֺЧֶ,BPĲѧϰ,ͳԽзѵ,ԴﵽܺõЧ.
Ȼͨʵʼ,ͳķѵBPȨֵܹȡýϺõЧ,ԻӾ,ӵľжʱȴ.Ϊ߷ѧϰĺ,ʱͬҪб,ͼͳļ㸴ԱȽϸ,ʹ߹ʵʱܵϴӰ.,ʹõͷֱͼΪģ񾭵,ռЧ.ͬʱ,򻯵ģЧ.
㷨ܽڡ塱ڼĲֹĻ.ȻȡһĽչ,ͬʱҲʶŲ,Ҫһƺ͸Ľ.,ӦڻӾϵͳķѵ,һо,ҵѷ㷨,֤ʵʱ.
߽ܣľ,1940,,ҪоΪ.
,1970,ʿ,ҪоΪӾ,ʵ.
ƽ,1958,ʿ,ҪоΪӾ.
ο,1936,,ҪоΪ,Զ.
,1935,,ʿʦ,йѧԺԺʿ,ҪоΪ˹,.
ͨѶϵˣľ, 100084,廪ѧܼϵͳصʵ
ߵλľӡڡƽοҡšࡡ廪ѧѧ뼼ϵ100084廪ѧܼϵͳصʵ 100084
ο
1Ii Hong Suh, Tae Wonkim. Fuzzy membership function based neural networks with applications to the visual serving of robot manipulators. IEEE Transactions on Fuzzy Systems, 1994,2(3):203220
2Liu Jian-zhuang, Xie Wei-xin. Pyramid segmentation of color images using fuzzy-C means cluster algorithm. In: Proceedings of 1993 IEEE Region 10 Conference Computer Commune Control Power Engine. 1993. 11301133
3Jang. ANFIS: adaptive-network-based fuzzy inference system. IEEE Transactions on Systems, Man and Cybernetics, 1993,23(3):665684
4Buckley J J, Yoichi Hayashi. Fuzzy neural networks: a survey. Fuzzy Sets and Systems. Amsterdam: North-Holland Publishing Company, 1994. 113
5Chao Y. Chen J, Teny C C. Simplification of fuzzy-neural system using similarity analysis. IEEE Transactions on Systems, Man and Cybernetics, 1996,26(2):180190
6Thomas Back. Evolutionary computation: comments on the history and current state. IEEE Transactions on Evolutionary Computation, April 1997,(1):317
1997-10-24յԭ,1998-01-09յ޸ĸ
