This is my response to a conversation between paul_virak_khuong and I off the list; my response included some questions to the list and I thought it would be nice if we could get feedback on some of the points from my response too. Cheers :-) RE: finger chords, you might remember a string of posts some years ago about finger chording recognition from video, and the resolution and frame rate was an issue discussed at the time, but I can't find any of those posts since the wearables.blu.org search feature seems to be broken currently. Is there a better way to search the list? If you're familiar with the tech behind head-tracking mouses and eye tracking I think you should do well in your research into this chording recognition stuff. I think it's also been globbed into discussions on this list and possibly elsewhere about gesturing, hand gestures, and object tracking in video. I do remember a keyboard device that watched the hands type on a virtual keyboard space that was defined by a projection of a 2D keyboard on a flat surface, the device used a laser and a camera to figure out finger movements. The device was aimed at executives with PDAs I think. Ah yeah, heres the link firefox favors in an address bar search(Im loving that, need to add wikipedia to that someday): http://www.alpern.org/weblog/stories/2003/01/09/projectionKeyboards.html I've been thinking about the way finger chords are seen in 2D space in a 640x480 pixel image, and here is how I see the challenge. In two parts, image processing, and then image analysis using computer vision techniques to abstract finger positions from the processed image. Firstly, the image processing might posterize the image sampled from the webcam to a certain extent (to be determined later), then, impose a new color map over the posterized image. Then I guess you make with the edge detection right? Im not sure if this is true of many webcams, but I have heard that a few have built-in image processing functions like posterize for example; does anyone have any idea what I might be talking about? Now that I think about it, uh, a color map rotation or substitution(what is the right word here?) would be pointless if the edge detection is done correctly, am I right? Secondly, the image analysis for determining finger positions from the processed image - I don't know page one about that so if anyone has any resources on computer vision of notable interest I would greatly appreciate any responses. Thankyou Paul, now I have an even greater anticipation for my copy of Intelligent Image Processing (Steve Mann, Hardcover) and my other new finds on amazon :-) Really can't wait to get my books :-)... never been more thrilled to get books than anything else at any time in my life until about the first of this year. Thankyou all, and good day. On 7/11/05, Paul-V Khuong <> wrote: > --- DLP <
> wrote: > > I've never really been able to use newsgroups with > > satisfaction(clients mostly) but now that I have > > Thunderbird installed > > and configured I will definitely see that I > > subscribe now to > > comp.lang.lisp. As for GCL, my attraction to it was > > the "as fast as C" > > claims and the FFI. SBCL and CMUCL are great and > > like you said much > > further ahead. > While GCL compiles down to C, that doesn't make it "as > fast as C". SBCL and CMUCL have much better optimisers > (given declarations on arguments and a couple > constructs Python can't infer through, speed is > usually within 0.9x to 2x as slow). >=20 > BTW, I might start working on recognising finger > chords in video, since we finally have small webcams > that do 640x480x32 fps. Do you have any references for that? >=20 >=20 >=20 > ____________________________________________________ > Sell on Yahoo! Auctions =BF no fees. Bid on great items. > http://auctions.yahoo.com/ > _______________________________________________ Wear-Hard mailing list
http://www.haven.org/mailman/listinfo/wear-hard
From Wear-Hard Mailing list Archive (WH)
Maintained by R. Paul McCarty
Archive created with babymail