You might try using IR. The IR absorption will be different for the liquid; H2O, for example, will have a high absorption band but the moist bubble may be different enough to use for detection. Just a thought. ----- Original Message ----- From: "Sarin Sukumar A" To: "Microcontroller discussion list - Public." Sent: Monday, May 17, 2010 1:17 PM Subject: Re: [EE] Computer Vision (Detection of circle or shape analysis) > Thank you for the advices. > in my application the target are bubbles, they are circles but slightly > deformed in shape also. Some comes in elliptical shape also; some are over > lapped also, this is a new information that convolution but is it like > cross-correlation? > i have tried that but it is not suitable if the objects are of different > scale. > if any one can give an idea i have a plan to make a new algorithm also. > i am looking for ideas; i have some but i want to discuss it with u people > YOurs SaRIn.... > > > On Mon, May 17, 2010 at 10:21 PM, Marcel Birthelmer > wrote: > >> On Mon, May 17, 2010 at 6:47 PM, Olin Lathrop >> wrote: >> > Marcel Birthelmer wrote: >> >> One way of detecting symmetric objects is convolution. >> > >> > There is no need for objects to be symmetric to detect their signature >> with >> > convolution. It does, however, alleviate orientation issues. >> > >> > >> >> If you don't know the shape of the object, but you do know that it's >> symmetric (which, granted, is not the case in this scenario), you can >> still use convolution. This is useful for example when detecting >> street signs in automotive vision applications, since pretty much all >> signs are symmetric in shape. >> >> -- >> http://www.piclist.com PIC/SX FAQ & list archive >> View/change your membership options at >> http://mailman.mit.edu/mailman/listinfo/piclist >> > -- > http://www.piclist.com PIC/SX FAQ & list archive > View/change your membership options at > http://mailman.mit.edu/mailman/listinfo/piclist > -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist