IT-LIST Digest 51Topics covered in this issue include:  1) Re: binary file question	by Laura Louise Cowen   2) Centroid calculation	by "Pascal Hubert"   3) !Help! Still Problem in DT3152 under Win95 	by Soonsung Hong   4) Re: overlapping discs radius calculation	by Bryan Crosby   5) Re: overlapping discs radius calculation	by (MD Squiers)  6) Re: identifying objects -Reply	by "Dr Pierre E. Ravussin"   7) Stacks	by Piero Bianco   8) Re: overlapping discs radius calculation	by Robert Smith   9) RE: Stacks	by "Brian O'Rourke"  10) FITS Format [NASA]	by Tom Mote ----------------------------------------------------------------------Date: Tue, 1 Oct 1996 19:16:28 -0700 (PDT)From: Laura Louise Cowen To: it-list@sparky.uthscsa.eduSubject: Re: binary file questionMessage-ID: <199610020216.TAA16364@fraser>> There is no mechanism for this in the current macro language.  The language> currently under development will provide this capability, you can assume a> 3-4 month wait time on the release of this version; however, we will make> available very (and I mean VERY) early interim releases in an effort to> expand our testing base.  Keep an eye on this list for further> announcements.> > Don> Sounds good for future use.  Is there a way to use the macro language toselect a section and always have it select the same size section nomatter what file it is selecting from.  I have got IT to do it on it'sown but it always uses the image used to create the file.  Let me knowwhen those interm release are made available.  Thanks, laura.> -- Laura Cowen      Indeed, it is an extreme irony of evolution that the only      creatures capable of uncovering their own evolutionary       history, choose so often to cover it up. -Robert Russell               ------------------------------Date: Wed, 2 Oct 1996 09:21:16 +0000From: "Pascal Hubert" To: it-list@sparky.uthscsa.eduSubject: Centroid calculationMessage-ID: <>Did you found the problem with the object centroid calculation ?Thanks,PascalPascal HubertForintek Canada Corp.2665 East Mall, Vancouver, BC, Canada, V6T 1W5Tel. (604)-222-5740FAX  (604)-222-5690e-mail Thu, 3 Oct 1996 00:19:31 -0400From: Soonsung Hong To: it-list@sparky.uthscsa.eduCc: hongsoon@egr.msu.eduSubject: !Help! Still Problem in DT3152 under Win95 Message-ID: <>Dear IT UsersI finally got the 'DTOL FG SDK for Win95' from Data Translation today.I installed both FG SDK for Win3.1 in \DTOL and for Win95 in \DTOL95,  and reconfigured DT3152 Device Driver to 60Hz, 1Mb memory.I have already checked the Plug'n'Play and Device Driver settings  in control pannel, and succesfully acquired several images with bundle softwares like Acq2Host and Passthru.But IT 1.25 still could not recognize my DT3152$0 at|Setting|Preference|DTOL Configure|Device| list.Did I miss anything important?Any suggestion and comment will be greatly appreciated!!!Soonsung Hong-----------------------------------------------------hongsoon@egr.msu.eduDept. of Materials Science and EngineeringMichigan State UniversityEast Lansing, MI 48824------------------------------Date: Thu, 3 Oct 1996 09:48:45 +0100From: Bryan Crosby To: it-list@sparky.uthscsa.eduSubject: Re: overlapping discs radius calculationMessage-ID: <>I will ask one more time in case anyone has some new ideas.I am looking to calculate the radii of the spheres or ellipses in animage. Sometimes the discs overlap, in which case it is calculationallypossible to calculate the radii of each disc contributing to anyidentified object. Clearly measuring the area of a object and gettingthe average radii by calculation is simple, but for ellipses is anaverage and for overlapping objects it is wrong.              So... , does any one have suggestions for availablesoftware, routines , algorithms, ideas on this knotty problem?Sincerely yoursBryan CrosbyPhysics & Astro,JCMB,University of Edinburgh.------------------------------Date: Thu, 3 Oct 1996 08:31:53 -0400From: (MD Squiers)To: it-list@sparky.uthscsa.eduSubject: Re: overlapping discs radius calculationMessage-ID: Oop, forgot to say what the Hough can do...It can map image pixels into a parameter space.  For example, in a searchfor lines, each pixel (of a binarized image) is tested to see if it fallson *any* line of interest.  Once all the pixels have been examined, theoutput of the algorithm is a matrix (for lines, more parameters requirehypercubes).  The peaks (which in themselves require some sorting out)indicate when the *likelihood* of a line exists.  In your case, a circlesearch will give you the locations and centers of the most likelycandidates for circles.  It should be able to detect a circle, even ifpartially occluded.  The hard task comes in filtering the output of thealgorithm...which is an area time currently doesn't permit me to explore.------------------------------Date: Thu, 03 Oct 1996 15:24:56 +0200From: "Dr Pierre E. Ravussin" To: it-list@sparky.uthscsa.eduSubject: Re: identifying objects -ReplyMessage-ID: <>At 11:44 16.09.96 +0600, you wrote:>Hi everyone,>>After some time, I bought a DT3155 for the Image Tool and now I have a lot>of problems and the computer crashes periodically as a result of trying to run>IT(1.25).>>I have a Pentium 100 MHz with 32 M RAM and am trying to capture images>from a monochrome camera at the moment and at the final stage, I*ll be>capturing images from a video player.>>I installed both DTOL and DTOL95,>I initially set it up for 1 M of Image RAM and I had problems trying to get>more than one image. So I increased it to 8 M and this still did not solve my>problem.>>For instance, >>while running the DT*s own software Acq2hst, it can only take 2 images and>then it returns with an unable to acquire status message of internal FIFO>overflowed (OLC_STS_FIFO_OVERFLOW), and then I can take another one>without this message and then every other capture gives the same error.>>When I try to run IT and select the source it performs an illegal operation and>shuts down.>>When I Acquire using Data Translation Acquisition: I get an Invalid page fault>in module CW3220.DLL and IT crashes.>>When I Acquire using Data Translation Multiimage Acquisition onto the stack,>the program crashes without getting any image giving an invalid page fault>(OWL50F.DLL and KERNEL32.DLL).>>After all these the windows become unstable and crashes.  >Is there anything I need to do to avoid these problems? And if so, what can I>do to get IT working with Image tool?>>Appreciate any comments.>>Thanks in advance,>>R.Naimi>I am using a dual Pentium 166MHz and have similar problems. It diminished alot when I upgraded to 64MB.RegardsPierre E. Ravussin ------------------------------Date: Thu, 3 Oct 1996 09:37:20 -0700 (PDT)From: Piero Bianco To: ImageTool Subject: StacksMessage-ID:  I'm having some problems saving stacks and I'm not sure exactly why.  Iopen a new stack, and then open images whcih are added automatically tothe stack.  When I save the stack and then re-open it, only the first twoimages are being saved. Can anyone please tell me what I am doing incorrectly?Thank you.Piero Bianco------------------------------Date: Thu, 03 Oct 1996 12:42:35 -0400From: Robert Smith To: it-list@sparky.uthscsa.eduSubject: Re: overlapping discs radius calculationMessage-ID: <>At 03:49 AM 10/3/96 +0600, you wrote:>I will ask one more time in case anyone has some new ideas.>I am looking to calculate the radii of the spheres or ellipses in an>image. Sometimes the discs overlap, in which case it is calculationally>possible to calculate the radii of each disc contributing to any>identified object. Clearly measuring the area of a object and getting>the average radii by calculation is simple, but for ellipses is an>average and for overlapping objects it is wrong.>              So... , does any one have suggestions for available>software, routines , algorithms, ideas on this knotty problem?>Sincerely yours>Bryan CrosbyBryan:  This ain't a trivial problem, but here's what I would do.    I seem to recall from previous letters that you can do a blob analysis onthis image, getting the separate blobs, along with their centres, boundingboxes, areas, perimeters. etc.  The ratio of area to perimeter is maximalfor a circle; those blobs which have this ratio are circles.  Done.    Those blobs which have smaller ratios may be composites or elipses.  Nexttake the centroid of the blob and find the major and minor axes.  You'llprobably have to do this by moving outward from the centre until you fallover the edge in all directions.  If the blob is an elipse, the axes will bewell-behaved; i.e. their 2 intersections with the edge will be equidistantfrom the centre in both directions. Also you can calculate the shape of theelipse from its axes, and check whether the measured area is correct for thegiven axes.  We now have the simple elipses and the composite figures remain.   Now "walk the perimeter" of the blob applying a 3-point curvature operator(The formula for this is well-known in analytic geometry, where "well-known"has the usual meaning -- I can't remember or lay my hands on it!)  This willgive you the intersection points of the overlapping figures because thecurvature will change sign at those points (i.e. convex changes briefly toconcave).  Now check the regions between intersection points to see if thecurvature is constant or changes smoothly.  This will differentiate circlesfrom elipses.  For the circular regions, find their centres (another"well'known formula") and if regions have the same centre, they're from thesame circle.  If you have enough of an elipse to get a maximum and minimumcurvature, yet another "well-known" formula will give you the equation.Failing that, you'll have to fall back on curve-fitting the regions whichappear to be parts of an elipse.  This is time-consuming, but hopefully itwon't happen often.  Well, I said it wasn't trivial, didn't I?  This should work well withnon-composite figures.  If they're composite, the S/N ratio will have to bepretty good, or the thing will unravel.  Curve-fitting is a more robustprocedure than 3-point curvature, but I've avoided using it because it's sotime-consuming.  If necessary, you could introduce it earlier, or use it torefine estimates given by curvature.  Given an estimate, you can limit thecurve-fitting to a narrow region of the parameter-space for an elipse, andreduce the search time considerably.Good luck!Bob             .  Robert A. Smith, Ph.D.  _____    .    Vision Systems' Analyst |     |<.      Current Technology, Inc. |_____|   .    (603) 868-2270     ^       .    / \   /   \------------------------------Date: Thu, 3 Oct 1996 13:09:01 -0400From: "Brian O'Rourke" To: "''" Subject: RE: StacksMessage-ID: <01BBB12C.0C2A5980@BOR(new)>------ =_NextPart_000_01BBB12C.0C338140Content-Type: text/plain; charset="us-ascii"Content-Transfer-Encoding: quoted-printableI have had similar problems with ver 1.25.  Don has posted a response =stating that he will look into the problem.  In the meantime, I have =been sucessfully saving stacks with IT Ver. 1.23.----------From: 	Piero Bianco[]Sent: 	Thursday, October 03, 1996 1:50 AMTo: 	bor@welchlink.welch.jhu.eduSubject: 	Stacks I'm having some problems saving stacks and I'm not sure exactly why.  Iopen a new stack, and then open images whcih are added automatically tothe stack.  When I save the stack and then re-open it, only the first =twoimages are being saved.=20Can anyone please tell me what I am doing incorrectly?Thank you.Piero Bianco------ =_NextPart_000_01BBB12C.0C338140Content-Type: application/ms-tnefContent-Transfer-Encoding: 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------ =_NextPart_000_01BBB12C.0C338140--------------------------------Date: Thu, 03 Oct 1996 16:57:58 -0700From: Tom Mote To: it-list@sparky.uthscsa.eduSubject: FITS Format [NASA]Message-ID: <>Has anyone attempted to open any of the NASA type FITS files?The Flexible Image Transport Format is widely used by NASA andastronomers.  I am an "it newby" and just attempted to open oneof my NASA files without success.  If any one has created amacro or knows some other way to open this kind of file, I wouldappreciate some assistance.Thank's,Tom------------------------------End of IT-LIST Digest 51