ARToolKit | Mailing List Archive |

Stable tracking with ARToolkit

From: | Rainer Blum <blum@i ......> | Received: | Jan 21, 2003 |

To | artoolkit@h .................. | ||

Subject: | Stable tracking with ARToolkit | ||

Hello everybody, I'm looking for some help concerning the stability of the tracking of ARToolKit. If a human being holds the pattern in his hand you have some kind of shivering effect since ATK processes (of course) every smallest movement. I heard that there are some references to source code solving this problem. Can someone provide me with these references or even with code solving this problem? Any help is appreciated! Rainer Blum ------------------------------------------------------------- Dipl.-Inf. (FH) Rainer Blum idmk - Institut für digitale Medien und Kommunikation Germany eMail: blum@i ...... |

From: | "Fivos DOGANIS" <Fivos.Doganis@i .................> | Received: | Jan 21, 2003 |

To | "Rainer Blum" <blum@i ......>, artoolkit@h .................. | ||

Subject: | RE: Stable tracking with ARToolkit | ||

Hello again, You can remove most of the jitter by using information from the previous frame, for pattern detection AND transformation computation. To make ATK use previous image information, use "arDetectMarker" instead of "arDetectMarkerLite" This latter makes ATK to look into the whole image in every frame. To compute the transformation matrix using the previously computed one (and therefore changing the value only if the difference between the two is important use "arGetTransMatCont" instead of "arGetTransMat". You have to store the previous position of the object by doing something like this: arGetTransMatCont(&marker_info[k], previous_trans, object[i].center, object[i].width, object[i].trans); previous_trans = object[i].trans; You must remember that the result is more stable, although less acurate, since the matrix is not recomputed in every frame! All information above has been found by looking at header files, unofficial documentation, informal documents etc. Therefore, the comments concerning the code are just assumptions. I don't know if everything is correct, but the result DOES look nicer. Hope this helps, Fivos DOGANIS -----Message d'origine----- De : owner-artoolkit@h .................. [mailto:owner-artoolkit@h ..................]De la part de Rainer Blum Envoyé : mardi 21 janvier 2003 10:42 À : artoolkit@h .................. Objet : Stable tracking with ARToolkit Hello everybody, I'm looking for some help concerning the stability of the tracking of ARToolKit. If a human being holds the pattern in his hand you have some kind of shivering effect since ATK processes (of course) every smallest movement. I heard that there are some references to source code solving this problem. Can someone provide me with these references or even with code solving this problem? Any help is appreciated! Rainer Blum ------------------------------------------------------------- Dipl.-Inf. (FH) Rainer Blum idmk - Institut für digitale Medien und Kommunikation Germany eMail: blum@i ...... |

From: | "Teis Draiby" <teis@d .........> | Received: | Jan 27, 2003 |

To | artoolkit@h .................. | ||

Subject: | Re: Stable tracking with ARToolkit | ||

It seems to me that the version of ARToolkit available from this site http://mtd.fh-hagenberg.at/depot/graphics/artoolkit/index.html tracks significantly more stable than the v. 2.52 I got from the ARToolKit Download Page. I only had a quick view, so I might be wrong. If you try it out I am curious to hear about your experiences. Teis > Hello everybody, > > I'm looking for some help concerning the > stability of the tracking of ARToolKit. > > If a human being holds the pattern in his hand > you have some kind of shivering effect since > ATK processes (of course) every smallest > movement. > > I heard that there are some references > to source code solving this problem. > > Can someone provide me with these references > or even with code solving this problem? > > Any help is appreciated! > > Rainer Blum |