英譯中3--有關(guān)服裝的,求高手
Figure 3: Garment data structure.2.3 Ontologies for garmentsWe can now introduce a class garments forsingle pieces of garment. Such a piece of gar-ment might be defined to be a collection of clothpatterns that are sewed together. This can beformally defined in languages like OWL thatprovide transitive properties. Commonly useddescriptions of garments such as jackets,trousers, skirts and dresses can nowbe introduced as sub-classes of garments.The propertyisPartOf ? cp × garmentsdenotes that a certain cloth pattern is part of acertain piece of garment.Using the ontology for cloth patterns it seemsto be possible to formally define these sub-classes of garments in good coincidence withthe common usage of the terminology trousers,skirts, etc. For instance trousers might be de-fined as a piece of garment that has a pattern Xfor whichisLyingOn(X,leftLeg)and has a pattern Y for whichisLyingOn(Y,rightLeg).Whereas a skirt might be defined as a piece ofgarment that has a pattern X for whichisLyingOn(X,bothLegs).However, in order to match the commonly usedterminology one might have to use negation, e.g.to state that for a skirt there is no pattern Zwhich is lying around the rightLeg, i.e.?isLyingOn(Z,rightLeg),as certain trousers also have patterns, for whichthe positive property given for skirts is true, too.Instead of trying to give formal definitions ofthese concepts it might be more useful to try tolearn these concepts from the relations given bytheir instances by supervised learning methods.2.4 Collections of garmentsThe virtual clothing usually is given by col-lection of garments, e.g. trousers, shirt, andjacket. The “world knowledge” that there is areasonable order in which to dress with the dif-ferent garments—e.g. that a jacket is above ashirt—can be formalized by a binary relationisDressedAfter on the garments.WecandefinearelatedpropertyisDressedAfterP on the level of sin-gle patterns. As a default implication we canstate that the property isDressedAfter(A,B)for two garments A and B implies the propertyisDressedAfterP(X,Y ) for all patterns X ofgarment A and all patterns Y of garment B.However, there are cases in which the prop-erty isDressedAfterP has to be defined dif-ferently, e.g. when a pullover is worn over a shirtbut the collar of the shirt should be worn over thepullover, cf. Sec. 3.1 and Figure 6.3 Applications3.1 Abstract modelingBy changing some values of the properties ofthe ontologies we have the possibility to modelvirtual clothing on an abstract level above thephysics-based modeling layer. For instance, theorder in which a virtual dressing of a collec-tion of garments is performed has severe con-sequences on the resulting geometry but can bespecified with our ontology by simply changingthe property isDressedAfter.Figure4showsthepre-positioningof trousers, which are given the nametrousersA, under a shirt—given the nameshirtB—onto a male human. In this case the4