о뷢չ
JOURNAL OF COMPUTER RESEARCH AND DEVELOPMENT
1999 36 9 Vol.36 No.9 1999



һ֪ʶĵ·ͼϵͳ
ɷҶ塡ΰ
ժҪ˵泵(ALV)ھϻ϶źܺõӦǰ.ΪALVļ֮һӾϵͳһֱܵע.ΪӾϵͳܣһ֪ʶĵ·ϵͳ.ϵͳв˶ϢںϺͶģƥĵ·ʶ𷽷.ַñϢڵĿռϢԺͶλľȷԣöѡԺõģⲻͬضԱȡ͸ٵĸţͬʱϢȶԣϾ֪ʶжεں.ʵԴû򼸺νṹӵĵ·ͼܸʶ.
ؼʡͼ⣬·ʶϢںϣ֪ʶϵͳ
ͼšTP391
A KNOWLEDGE-BASED ROAD SCENE UNDERSTANDING SYSTEM
FAN Cheng-Fa, YE Xiu-Qing, and Gu Wei-Kang
(Department of Information and Electronics Engineering, Zhejiang University, Hangzhou 310027)
AbstractALV (autonomous land vehicle) has a alluring prospect both in military and civilian applications. As the one of the kernel techniques in ALV, vision system has received close attention in the past two decades. In the paper here, a road recognition algorithm combining fusing of multiple information and matching with multiple models, in a knowledge-based system KRUS is presented, with high performance achieved. In the algorithm, a unique edge detection method is developed, from which the benefit of good localization is received, while the disturbance of noise is reduced. Then the robust region information is fused with edge information, with the help of some knowledge of the scenes. The experiment results show that most of the ill-formed or complicate-structured images are very satisfying.
Key wordsimage understanding, road recognition, information fusing, knowledge-based system
1
80һALV (autonomous land vehicle) ʾɹ˺о̵룬ԼӲĽٶȸ졢ܸǿALVҲ½֣16. ALVϵͳУӾϵͳһҪɲ֣ҲӰALVٶȺԵĹؼ.Ŀǰ£ʹϵͳȫ⣬һЩ㷨һЩоҲǳҪ.ΪȻǳӣ㷨ģ޷Ӧֱ仯ǹһ֪ʶĵ·ͼϵͳKRUS (a knowledge-based road scene understanding system) ϣֻ֪ͨʶһӦԺ;ȷԶܺõϵͳ.
ӦԺ;ȷǺһӾϵͳӵָ.·ʽֶ·ϵľ۸ӣ绷ǧ.һʵõALVӾϵͳҪԲͬ·ͺ绷õӦ.ȷ㺬壬һҪõȷĵ·λãҪȷ֪·˽ṹת䡢Ϣ.
2ϵͳ˼
ͼĿʶ㷨ݽǶȵĲͬϿɷΪ򷽷ͱط.͵򷽷صɫʡȻһЩƵΪһռ䣬Ȼ㷨ȷͼÿһĿ껹ǷĿ.ַŵǶ̫УʹǿӰҲܵõȷ·.ҪȱԻ·ߵľȷλϢ.
ڱص㷨Ӧõ·ʶϣע·仯ƽ··֮ͻʵڷӳͻıصͼУѰ·.󲿷ֵ·ǰͼıеģñʾ·Ϊȷ.һ棬Ϣһ־ֲϢҪ跨ڰָűصıͼȡĵ·.
Ϣڵȫֲֲ뾫ȷĻԣϢںϵķǳ.ΪñϢĶλȷԺϢӦԣԱϢΪͨϢصĲֲȷԣϾ֪ʶںϷͬʱӦԺ;ȷָϻõܣϵͳ㷨ƵҪָ˼.
ϵͳƻѭԭ.ԭ֮һǲø֪ʶģ.Ե·ĸԣܽøֹڵ·֪ʶٲȷԣڹģʱ뾡ǿĸ.ԭ֮ǼһЩɳڵĻ.㷨ģ֮һЩɳڣڶģ֮ЭԱ㼰ʱһЩ©ԸϵͳȶԷǳҪ.
3ϵͳ·ʶ㷨
3.1㷨
Ӷάɫͼʶ·ȽϽӽʵļ.⼸ĻϣͿҵϵͳʵֵֵ.
1. ·ȸ·߱.
ҪΪ˼.
2. ·ı߱ȽϹ⻬.
·һ˹ṹһձãһЩͻζļνṹ.
3. ǰԶľһǵ·.
ʻʱ.
3Ļϣǹ˷ϵͳ˼㷨ṹͼ1ʾ.һЩͼƬͼ2.

ͼ1ϵͳ·ʶ㷨ͼ

(a)(b)(c)(d)

(e)(f)(g)(h)

(i)(j)(k)(l)
(a) ·ԭʼͼG (b) ͼ (c) 򷽷· (d), (e), (f), (g) L, R, H, Vǿͼ (h) LL or RL (Ϊ˼ʾͼƬ LLRLֵͼ) (i) HLL or HLH or VLL or VLH; (j) LB or RB, LG or RG (ͼ£֪䣬Էһʾ) (k) HBL or HBH or VBL or VBH; (l) ʶ·.
ͼ2ϵͳ·ʶ㷨һЩͼƬ
3.2߼
ʹƥܺáǿķģ壬߱߼.ϵͳв4ģ壺ߡұߡˮƽߺֱģ(ͼ1).ģĺ˲ø˹ģ屻֤ǽӽܣ7. 裬ȡͼϵзΪxзΪyͼΪI(x,y)ؼΪE(x,y)ôұߵۼ㹫ʽΪ

(1)
ֱߵۼ㹫ʽΪ

(2)
ˮƽֱƣx,y.ʵʵֵУ1111Сģ壬е.⼸ģɵõLE(left edge), RE, HEVE 4ıǿͼ(ͼ2(d)(g)).
ģõѡԣ˷ƥصӦرǿƷױӰصӰ.һȡҲ˶.صǿͼy֣ҵ.ȻCannyıȡ˫ֵ7. ıǿڸֵΪοص.ڵֵĵΪѡص㣬ͬʱںѡصнзѡԵҵһ.г̵̫ģԼȱ㹻ֵοصֵ֧õLL (left lines). ƵصõRL Ϸȵ·HLL (horizontal lines with lower upward luminance), ෴HLHԼֱVLLVLH(ͼ2(h),(i)).ЩȡУֵȡýϵ֤ͣҪ·Ϊһ壬ʵ£ģ뷴.
3.3Ϣȡ
ϵͳУѡͼȷֲӦĲɫͷɾֲϢϢԲɫӦ㷨ͼдԷָõĽ(ͼ1).ȷֲϢԶͼG(ͼȾнϺõĶӦϵ)ֱͼķ.ӦĲɫԴߵһƪ£8ݵ뷨һЩĽ.򵥵˵ڼ3Ļϣѡһ·ͳϵR-GB-GάֱͼȻʽΪR-GB-GɫʷֲΪ·Ĳɫ.ԶȻԵļ9ݣͼصΪf(x,y),ƽHolderɱΪ

(3)
УSΪx,yĦ.(x,y)뾰ķάйأֱӷӳ˾.ʵӦѡ񦤦=2Ϊ.
3.4߹
ЩϢ͵ģ飬жεں.ȣϢһЩ֪ʶԸ߽йˣȥŵ(ͼ1).ߣһЩ֪ʶԴз.
 һ󷽺ҷͬʱ··ҷƽȽϸߣҲѷ·ôΪһ·м˳.
 д·Ⱥܵ͵ת䣬ҷ·ô·ѷ·磬ںһж.
֪ʶԴ࣬ϸµضԱ߽й.ںϹ֮󣬵õһЩѡĵ·ߣΪLB(left boundaries), RB, HBL, HBH, VBLVBH(ͼ2(j),ͼ2(k)).
3.5ұ߸֪
ߺұ߶Ե·ǳҪҪұ߽и֪飬.֮һѰ.߶·ߵѰȷҪ˴֪ʶӸۺϿǣʹ˹˺LBRBṩݴҲοǰLLRL(ͼ1).֮Ǵ߿ʼĸ֪ԤԼϢľϲݵȥݵļʧܵ.֪֮õһЩ飬µϵ֮ϣһľ.Ҳ˵֮ͬȱڣЩȱڿֻ·ȱʧģҲζһֲ棬Ե·ṹжǺҪ.
3.6·ģƥ
֪ĽLG(left line groups)RGʾ(ͼ2(j)). LGRGԼHBLHBHVBLVBHȱϢ;ֲϢһ͵ģƥģУе·ģ͵ƥͱߵ.
·˽ṹʽֶԲɴֵϸ󾫵ķʽ׼ȷƥ.ȣLGRGĿת˫ߡ󵥱ߡҵ߻Ч4ͬģ飬4ʵ϶ӦĴģ.ÿһģڣнһƥж.ģƥȷڵǰģнһ󾫣ֲҳԭתΪģƥ䡢жϺ.磬һЩ֪ʶԴָģһЩת.
 (ģ)ӳͼϱߵĽͼϱұһ룬ûзHBLϵĳͬʱԵߴڣұ߸RGת˫ģ.
 (˫ģ)ֻʣҵߣתҵģ.
֪ʶԴʹģƥ価׼ȷ.ȷ˴ģʽ֮󣬽һ.󾫹УֲмֲЩ˽ṹϸڱⲢһЩ.󾫹ҲӦйؾ֪ʶ֮·ƣHBLHBHVBLVBHҵߵ֤ݣôΪһм佻棬ҳ׼ȷ.·ʶһ·߱պʽ(ͼ2(l)).
4ʵչ
Ǵ100ͼû˽ṹӵͼѡ˼ʮ͵ͼڷЩͼĻϣϵͳĵ·ʶ㷨.֮ѡ300ͼ˲ԣȡԤڵõĲͼȷʶ𣬶Դõͼ(200)ܸȷʶ.Ըӵĵ·˽ṹȷ֪.ͼ3(a)(b)ͼµʶǿӰ·ˮ£ȷҳ·.ͼ3(c)(d)˶Էֲתȷ֪.ϵͳʵֵף10м.δϵͳ㷨һŬͬʱĿʶģ飬ͨMMXٺͶദеİ취ʵʵʱ.


ͼ3һЩ͵ĵ·ͼʶ
ע:õƹί塱شƼĿʽӾϵͳ.
߼:ɷУʿ߼ʦҪоΪͨż.
Ҷ壬ŮڣҪоΪͼӾ.
ΰ.УڣʿʦҪоΪܿơӾ.
ߵλ:㽭ѧŵϵݡ310027
ο
1Thorpe C et al. Toward autonomous driving: the CMU navlab, part I-perception, IEEE Expert, 1991,(8)
2Crisman J D, Thorpe C E. SCARF: A color vision system that tracks road and intersections. IEEE Trans on Robotics and Automation, 1993, 9(1)
3Denasi S, Quaglia G. Line grouping using perceptual saliency and structure prediction for car detection in traffic scenes. SPIE-Intelligent Robots and Computer Vision XII, 1993, 2055
4Maurer M et al. VaMoRs-P, and advanced platform for visual autonomous road vehicle guidance, SPIE-Mobile Robots, 1994, 2352
5Broggi A. Parallel and local feature extraction: A real-time approach to road boundary detection. IEEE Trans on Image Processing, 1995, 4(2)
6Gracfe V. Research on robot vision and intelligent robots: Approach, result, publication. Measurement Science, Federal Armed Forces University, Munich, 1997
7Canny J, A computational approach to edge detection. IEEE Trans on PAMI, 1988, 8(6)
8Fan C et al. An adaptive method of color road segmentation. In: Proceedings of CAIP'97. Kiel, Germany, 1997
9Chen S S et al. Shape from fractal geometry. Artificial Intelligence, 1990, 43
10ɷ. һӻͼ. йͼͼѧ, 1998,3(9)
(Fan Chengfa et al. A visual and integrated image processing environment. Journal of Chinese Image and Graphics, 1998, 3(9))
ԭյڣ1997-12-09޸ĸյڣ1998-05-05.
