hello Anthony, you can set some parameters when using the library, to get what you want. For example:
- you can set what will be the maximum size of a blob. If the blob is greater than the maximum size, the algorithm checks if there is really a large blob, or several small blobs. If the blob is smaller them maximum size, the algorithm stops there. Of course, smaller the maximum size, slower the blob detection will run; - you can set if you want edges or just the bounding rectangles; So, the library can see fingers as several blobs, or as just one blob, based on parameters given. I've also written some classes to make blob tracking, but I haven't published them in my blog yet because I want to write some docs and examples. But I think they can be useful for you, so you can grab them here: http://www.andreithomaz.com/arquivos/blob_tracking.zip best, andrei On Thu, Jan 8, 2009 at 12:47 PM, Anthony Pace <[email protected]>wrote: > great info... > > Before I peruse through your library, I have a few questions. Do you test > for the largest circle or rectangle that can fit in the blob? Or for the > centre point of the blob? or do you look for a circle or rectangle that > encapsulates the blob, and then possibly locate an approximate centre. > > I would think just locating the centre point of a rectangle or circle would > be faster to track than looking for edges. What happens when the fingers > are placed together? do you count them as one finger and then add a size > restriction? Do you restrict actions based on the size of a blob? if so how > has that worked out? > > I am definitely going to pick through it; yet, I would love some insight. > > > Andrei Thomaz wrote: > >> and, of course, don't miss the NUI group forum: >> http://nuigroup.com/forums/viewforum/33/ >> >> and the original version of blobDetection >> http://v3ga.net/processing/BlobDetection/ >> >> >> best, >> andrei >> >> >> On Thu, Jan 8, 2009 at 12:10 PM, Andrei Thomaz <[email protected] >> >wrote: >> >> >> >>> hello, >>> I've ported a blob detection Processing library, blobDetection, to AS3. >>> It >>> is slow, but maybe you can optimise it. >>> http://andreithomaz.com/labs/?p=8 >>> >>> Using BitmapData.getBoundsRect(), I started BitmapBlobDetection. It works >>> faster: >>> http://andreithomaz.com/labs/?cat=4 >>> >>> I've also developed a kind of register of visual objects to be affected >>> by >>> blobs movements. I need to make some examples before publishing it, but >>> feel >>> free to ask me the files privately. I have used the register in the >>> touchscreen jukebox (http://andreithomaz.com/labs/?p=21). >>> >>> >>> best, >>> andrei >>> >>> >>> >>> On Thu, Jan 8, 2009 at 8:58 AM, Piers Cowburn <[email protected]> >>> wrote: >>> >>> >>> >>>> There's good stuff by Grant Skinner too: >>>> http://incomplet.gskinner.com/index2.html#camwriter >>>> >>>> Piers >>>> >>>> >>>> >>>> On 8 Jan 2009, at 10:35, Ian Thomas wrote: >>>> >>>> Yeah, I absolutely get the idea (as I said, we were discussing it >>>> >>>> >>>>> yesterday - with reference to an infra-red camera, but same >>>>> principle). Check out Mario's stuff, there are areas-of-colour >>>>> recognition algorithms there which will be of help, I think. >>>>> >>>>> Cheers, >>>>> Ian >>>>> >>>>> On Thu, Jan 8, 2009 at 10:08 AM, Anthony Pace < >>>>> [email protected]> >>>>> wrote: >>>>> >>>>> >>>>> >>>>>> To my knowledge all flash integration for multi-touch has been made >>>>>> possible >>>>>> by using flosc by ben chun, or with OpenCV and touchLib; however, what >>>>>> I >>>>>> am >>>>>> proposing is that it is possible directly through the use of the flash >>>>>> player and a simple cam rig. I think that it should be possible, and >>>>>> at >>>>>> least somewhat comparably fast, considering the player doesn't have to >>>>>> parse >>>>>> xml socket data on a continuous basis, and all image conversions are >>>>>> done by >>>>>> the player. I do concede that doing in c++ will most likely prove to >>>>>> be >>>>>> faster; however, I figure why not try it out. >>>>>> >>>>>> Since I know c++ somewhat I figure I will probably look at their >>>>>> libraries >>>>>> if I get stuck; yet, in the meanwhile, this is a nifty experiment. >>>>>> >>>>>> Here is the video that sparked it... >>>>>> http://www.youtube.com/watch?v=pQpr3W-YmcQ >>>>>> >>>>>> Colour tracking would be efficient in a system where you have multi >>>>>> coloured >>>>>> thimbles or reflective tape, which is what I was thinking of when I >>>>>> came >>>>>> up >>>>>> with this idea. >>>>>> >>>>>> >>>>>> http://anthony-pace.deviantart.com/art/add-VR-to-current-FPS-GAMES-91091960 >>>>>> >>>>>> Ian Thomas wrote: >>>>>> >>>>>> >>>>>> >>>>>>> I was actually thinking of Mario's lecture at FoTB 2008 where he used >>>>>>> very similar algorithms to read square barcodes (whatever those >>>>>>> things >>>>>>> are called) from a webcam. Not multitouch, but there's a lot of >>>>>>> similar code (spotting areas of colour, eliminating noise). >>>>>>> >>>>>>> Worth a look, Anthony. >>>>>>> >>>>>>> Funnily enough we were talking about exactly the same system for >>>>>>> multitouch here yesterday - spotting areas of colour. One of our >>>>>>> developers had seen some demo of a similar system on TV over the >>>>>>> weekend... >>>>>>> >>>>>>> Ian >>>>>>> >>>>>>> On Thu, Jan 8, 2009 at 9:12 AM, Glen Pike <[email protected] >>>>>>> > >>>>>>> wrote: >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>>> Hi, >>>>>>>> >>>>>>>> Mario Klingemann demo'd something like this at FlashOnTheBeach 2007 >>>>>>>> - >>>>>>>> have >>>>>>>> a look at quasimondo.com to see the "2D or not 2D lecture" - not >>>>>>>> sure >>>>>>>> if >>>>>>>> he >>>>>>>> released any code though? >>>>>>>> >>>>>>>> http://lectures.quasimondo.com/ >>>>>>>> >>>>>>>> Glen >>>>>>>> >>>>>>>> Anthony Pace wrote: >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>>> I have figured out how to make flash work as a multi-touch >>>>>>>>> system... >>>>>>>>> I >>>>>>>>> am >>>>>>>>> coding it now... >>>>>>>>> >>>>>>>>> it is so darn simple... >>>>>>>>> it just came to me after watching a video on how to make your own >>>>>>>>> multi-touch input device using a webcam a piece of white paper and >>>>>>>>> a >>>>>>>>> box >>>>>>>>> to >>>>>>>>> block most of the light so the cam only sees light and dark. >>>>>>>>> >>>>>>>>> Since it is all based on blobs, and processing the image through a >>>>>>>>> filter >>>>>>>>> to break down the logic, and flash has access to web cams, I can >>>>>>>>> just >>>>>>>>> convert the image to black and white, threshold the image to find >>>>>>>>> the >>>>>>>>> blobs, >>>>>>>>> average out a hit test area based on the finger positions, and all >>>>>>>>> that >>>>>>>>> would come after is putting together the finger gesturing logic. >>>>>>>>> >>>>>>>>> for instance if two blobs exist around their averaged points and >>>>>>>>> have >>>>>>>>> not >>>>>>>>> moved too much or at all, yet there is a third blob that is moving >>>>>>>>> a >>>>>>>>> significant amount, perform a specific function based on its >>>>>>>>> direction >>>>>>>>> (directional logic through tracking input fed into basic physics >>>>>>>>> equations >>>>>>>>> and Pythagorean concepts) >>>>>>>>> >>>>>>>>> If someone has beaten me to it, I would love to know; however, if >>>>>>>>> you >>>>>>>>> beat >>>>>>>>> me to it, after reading my post, tell me about it and show me your >>>>>>>>> source so >>>>>>>>> my life is made easier. >>>>>>>>> _______________________________________________ >>>>>>>>> Flashcoders mailing list >>>>>>>>> [email protected] >>>>>>>>> http://chattyfig.figleaf.com/mailman/listinfo/flashcoders >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>> _______________________________________________ >>>>>>>> Flashcoders mailing list >>>>>>>> [email protected] >>>>>>>> http://chattyfig.figleaf.com/mailman/listinfo/flashcoders >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>> _______________________________________________ >>>>>>> Flashcoders mailing list >>>>>>> [email protected] >>>>>>> http://chattyfig.figleaf.com/mailman/listinfo/flashcoders >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>> _______________________________________________ >>>>>> Flashcoders mailing list >>>>>> [email protected] >>>>>> http://chattyfig.figleaf.com/mailman/listinfo/flashcoders >>>>>> >>>>>> _______________________________________________ >>>>>> >>>>>> >>>>> Flashcoders mailing list >>>>> [email protected] >>>>> http://chattyfig.figleaf.com/mailman/listinfo/flashcoders >>>>> >>>>> >>>>> >>>> _______________________________________________ >>>> Flashcoders mailing list >>>> [email protected] >>>> http://chattyfig.figleaf.com/mailman/listinfo/flashcoders >>>> >>>> >>>> >>> >>> >> _______________________________________________ >> Flashcoders mailing list >> [email protected] >> http://chattyfig.figleaf.com/mailman/listinfo/flashcoders >> >> >> > > _______________________________________________ > Flashcoders mailing list > [email protected] > http://chattyfig.figleaf.com/mailman/listinfo/flashcoders > _______________________________________________ Flashcoders mailing list [email protected] http://chattyfig.figleaf.com/mailman/listinfo/flashcoders
