IT-LIST Digest 26Topics covered in this issue include:  1) Specialized X,Y-Coordinate Value Output	by "Regis the Watcher"   2) Re: Specialized X,Y-Coordinate Value Output	by Robert Smith   3) IT won't install	by Ron DeSpain   4) Re: Specialized X,Y-Coordinate Value Output	by "S. Brent Dove"   5) RE: trace measure	by Don Wilcox   6) RE: Counting black and white pixels	by Don Wilcox   7) RE: Virtual Memory Problem	by Don Wilcox   8) Re: Specialized X,Y-Coordinate Value Output	by "Peter D. Cherry"   9) Plug-In development using visual C++	by "WANG, XIAOHAN"  10) Re: Specialized X,Y-Coordinate Value Output	by "Dr. Norman Capra" ----------------------------------------------------------------------Date:    Wed, 26 Jun 96 14:49 EDTFrom: "Regis the Watcher" To: IT-LIST@sparky.uthscsa.eduSubject: Specialized X,Y-Coordinate Value OutputMessage-ID: <9606261935.AA09398@sparky.uthscsa.edu>Our laboratory is analyzing morphological variation in posterior lobe shape inhybrid Drosophila species.  This structure is distinct for each species and thehybrids show a blending of parental types.  Our goal is to quantify both size and shape aspects of lobe variation through the use of a morphometric descriptor based on elliptical Fourier and principal components analyses.  This will allow us to to obtain a small number of variables that encapsulate much of the information about variation in shape.In order to do this we need to process .TIF files of our images in the following way:1) Convert the original black-and-white image to an outline of one-pixel width2) Obtain the X,Y-coordinate values of each pixel in the outline IN A MANNER WHICH FOLLOWS THE PATH OF THE CURVE.  In other words, a sequential list of values from least to greatest will not work; the coordinates must be written to filein the order in which they would occur if a pencil were to draw the curve in time.We have found that we can approximate this procedure by first applying thresholding and filtering to the image, and then using the POINTS function to manually plot the X,Y values and write them to a file.  However, this system is time-consuming, offers limited resolution of the curve, and is dependent on the accuracy of the individual who plots the points.  A system that could internally generate a continuous set of coordinate values for every pixel in a smooth curve of one-pixel width as the curve is swept out in time would provide for the best possible input for mathematical Fourier analysis.Is this procedure possible using Image Tool, and if not can anyone recommend amore effective method than the one we are using?  Any help or information would be greatly appreciated.  Thanks.Michael AbrahamPenn State Biology Department208 Mueller LabUniversity Park, PA 16802(814)-865-2715mta108@psuvm.psu.edu------------------------------Date: Wed, 26 Jun 1996 16:16:08 -0400From: Robert Smith To: it-list@sparky.uthscsa.eduSubject: Re: Specialized X,Y-Coordinate Value OutputMessage-ID: <1.5.4.32.19960626201608.0067913c@curtech.com>At 02:44 PM 6/26/96 +0600, you wrote:>>1) Convert the original black-and-white image to an outline of one-pixel width>2) Obtain the X,Y-coordinate values of each pixel in the outline IN A MANNER WH>ICH FOLLOWS THE PATH OF THE CURVE.  In other words, a sequential list of values> from least to greatest will not work; the coordinates must be written to file>in the order in which they would occur if a pencil were to draw the curve in ti>me.Sounds like you want to build a "line-walker" Plug-in. This scanns the 1-bitimage row-by-row, until it strikes a line.  It then walks around that lineto an end, turns around, and outputs all the points as it walks back overthe whole line, erasing points as it goes.  There are a few subtleties(closed curves, forks, ..) but these aren't very difficult to handle.  Whenyou've scanned the whole image all the points will have been listed.  Themain problem is whether your pre-processing gives you ALL the points, or arethere some gaps.  Most line-walkers keep an eye on the original image, usingit to fill in lines/edges which fell below threshold and left a gap.    This isn't a real difficult algorithm to write and -- unless you do hairythings to try to speed it up -- should occupy only a few pages of code.Good luckBob             .  Robert A. Smith, Ph.D.  _____    .    Vision Systems' Analyst |     |<.      Current Technology, Inc. |_____|   .    (603) 868-2270     ^       .  ras@curtech.com    / \   /   \------------------------------Date: Thu, 27 Jun 1996 18:19:13 -0700From: Ron DeSpain To: "IT-LIST@sparky.uthscsa.edu" Subject: IT won't installMessage-ID: <31D33311.6D51@ez2.net>I downloaded IT v 1.23 and the DLL, unzipped IT and ran setup.exe only to be presented with a message that read "Setup requires _setup.dll & _isres.dll located in _setup.lib.   _setup.lib must be located in tha same directory as setup"   _setup.lib is in the directory, but the instalation will not proceed.   Can anyone help?Thanks-- Ron DeSpain - Chief EngineerCostar Precision Optics & Video Products40925 County Center Dr., # 210Temecula, CA 92591909 699 9000------------------------------Date: Fri, 28 Jun 1996 08:00:53 -0500 (CDT)From: "S. Brent Dove" To: IT List Subject: Re: Specialized X,Y-Coordinate Value OutputMessage-ID: <01I6FO3H74I6007Q4U@uthscsa.edu>Recently on ITLIST>In order to do this we need to process .TIF files of our images in the >following way:>1) Convert the original black-and-white image to an outline of one-pixel width>2) Obtain the X,Y-coordinate values of each pixel in the outline IN A >MANNER WHICH FOLLOWS THE PATH OF THE CURVE.  In other words, a sequential list of >values from least to greatest will not work; the coordinates must be written to >file in the order in which they would occur if a pencil were to draw the curve >in time.>Is this procedure possible using Image Tool, and if not can anyone recommend a>more effective method than the one we are using?  Any help or information >would be greatly appreciated.  Thanks.As you have already found out ImageTool creates an outline of all objects that are found using the FIND OBJECT in the ANALYSIS menu.  It would be possible to write a IT Plugin that would output the coordinates of each object in any order that you wish.  At present the FIND OBJECTS does not perform this feature.  If anyone else could use this feature please send information about applications to ITLIST.So long for nowS. Brent Dove                            Voice: (210) 567-3333Diagnostic Sciences                      Fax:   (210) 567-3334University of Texas                      Email: dove@uthscsa.eduHealth Science Center                    Web:   ddsdx.uthscsa.eduSan Antonio, TX    USA                   ftp:   maxrad6.uthscsa.edu------------------------------Date: Thu, 27 Jun 1996 14:07:02 -0700From: Don Wilcox To: "'ImageTool List Server'" Subject: RE: trace measureMessage-ID: <01BB64C0.3F6FA380@slip15.xroads.com>----------From:  James Zheng[SMTP:zhengjq@umdnj.edu]Sent:  Monday, April 29, 1996 12:29 AMTo:  wilcox@xroads.comSubject:  trace measureHello,I am trying to use IT to do some quantitative measurements of nerve =3Dcells in microscopic images.  One thing is to measure the trajectory of ==3Da nerve process.  I found out that IT can not do a trace measure, in =3Dwhich case you click and hold down left mouse button to trace the =3Dtrajectory.  The "distance" function certainly can do it but is very =3Dinconvenient.  Is IT going to add this feature or Is there a plug-in for ==3Dthis?  It should be also nice if one can trace-measure an area.  It =3Dwould be better if one can set the interval between pixels during =3Dtrace-measure. =3D20The distance tool can measure multi-segment lines.  To do this, click =and release the mouse on the endpoint of the trajectory, and the click =and release on successive points along the line, double-clicking on the =last point to get the line's length.  You can use the Distance tab in =the settings dialog to select the measurements to be produced.Don------------------------------Date: Thu, 27 Jun 1996 13:48:44 -0700From: Don Wilcox To: "'ImageTool List Server'" Subject: RE: Counting black and white pixelsMessage-ID: <01BB64C0.2EE6D9C0@slip15.xroads.com>----------From:  Emily B. Kim[SMTP:ebkim@dcn.davis.ca.us]Sent:  Thursday, June 20, 1996 11:02 AMTo:  wilcox@xroads.comSubject:  Counting black and white pixelsi have a picture that's black and white.  i basically need to find a =ratioof black to white pixels so that i can compare a bunch of these pics.  =inphotoshop, the histogram will show you the total number of pixels, plus, =ifyou point the mouse over the left and right side of the graph, it will =giveyou the number of pixels that are black and the number that are white,respectively (no grays --- 2-bit graphic).  this information really =needs togo into a spreadsheet of sorts.  PICTURE 1:  column A --- black pixels;column B --- white pixels; column C --- total pixels; column D --- ratio =ofblack pixels to white pixels.  the same for pictures 2, 3, etc.  when i =usethe imagetool program and try to use "count black and white pixels" itdoesn't give me those numbers that i mentioned above.  it gives me means =andstandard deviations, and i can't figure out on what #'s it's basing =thesefigures.  sigh.  perhaps i am misunderstanding what "count black and =whitepixels" is doing.  can someone please help me?  thanks!  -emilyThe count black/white pixels plug-in was developed by me, the ImageTool =developer.  The plug-in operates on 2-color grayscale images, produced =by the manual thresholding process.  To use the plug-in, copy it into =the Plug-Ins directory along with the other plug-ins.  The plug-in =provides a settings sheet that you can use to configure the statistics =you want.  The process is as follows:1.  Produce a 2 color image.2.  Select Count Black/White Pixels from the Analysis menu.3.  The program will automatically count the pixels in the two =categories, and add a line to the results window.  Note that when =numbers are added to the results window, the ImageTool program =automatically produces mean and standard deviation statistic for each =numeric column, so the count of black and white pixels will start in the =third row from the top.If this does not help, let me know.Don------------------------------Date: Thu, 27 Jun 1996 14:00:00 -0700From: Don Wilcox To: "'ImageTool List Server'" Subject: RE: Virtual Memory ProblemMessage-ID: <01BB64C0.38A8B6E0@slip15.xroads.com>----------From:  Earl J. Sanford 4-6358[SMTP:SANFORD_EJ@corning.com]Sent:  Friday, June 14, 1996 1:24 AMTo:  wilcox@xroads.comSubject:  Virtual Memory Problem      	Hello, I'm another new user to Image Tools, version 1.23. I am=20      currently experiencing a problem that I have been unable to=20      resolve. I am trying to size some objects in a 1024 x 1024 TIF=20      image. I am successful at creating a binary image using either=20      manual or automatic thresholding. I run into a problem when I=20      execute the find objects command. The problem is my computer comes =      back with a error statement stating that I am running low on=20      virtual memory, please close other applications .....      	I get this error even when I am running only image tools. The=20      computer I am using has an INTEL 120 MHz pentium processor with 16 =      megs of RAM and a 1 Gig hard-drive and is operating under Windows=20      NT. I have tried to increase my virtual memory, as high as 200=20      megs, but this does not seem to resolve my problem.      	I would greatly appreciate it if anyone can help me with this=20      problem. If you can, please feel free to contact me in any way you =      feel comfortable. My snail-mail, e-mail, and phone numbers are=20      listed below. Additionally, I can send electronically a copy of =one=20      of my images to anyone who wants one if that would help.      	I am also wondering if ImageTool is a true 32-bit application or=20      is it a 16-bit application that can be run by Windows NT.Sorry for the delay.  The problem is most likely related to an oversight =in the Find Objects code that does not detect images that have a very =large (> 1000) number of small objects, like in a noisy image.  The =result is that the algorithm uses a lot of memory in these cases, when =in fact it ought to complain and exit.  I will be addressing this in the =next release of the program.As for the nature of ImageTool, it is a true 32 bit program developed =specifically for Windows 95 and Windows NT 3.51.  In fact, it will not =run under Windows 3.11, even using Win32s.Don------------------------------Date: Fri, 28 Jun 1996 16:04:12 -0400From: "Peter D. Cherry" To: it-list@sparky.uthscsa.eduSubject: Re: Specialized X,Y-Coordinate Value OutputMessage-ID: <31D43ABC.1E69@scisun.sci.ccny.cuny.edu>S. Brent Dove wrote:> > Recently on ITLIST> > >In order to do this we need to process .TIF files of our images in the> >following way:> >1) Convert the original black-and-white image to an outline of one-pixel width> >2) Obtain the X,Y-coordinate values of each pixel in the outline IN A> >MANNER WHICH FOLLOWS THE PATH OF THE CURVE.  In other words, a sequential> list of> >values from least to greatest will not work; the coordinates must be written> to> >file in the order in which they would occur if a pencil were to draw the> curve> >in time.> > >Is this procedure possible using Image Tool, and if not can anyone recommend a> >more effective method than the one we are using?  Any help or information> >would be greatly appreciated.  Thanks.> > As you have already found out ImageTool creates an outline of all objects> that are found using the FIND OBJECT in the ANALYSIS menu.  It would be> possible to write a IT Plugin that would output the coordinates of each> object in any order that you wish.  At present the FIND OBJECTS does not> perform this feature.  If anyone else could use this feature please send> information about applications to ITLIST.> > So long for now> > S. Brent Dove                            Voice: (210) 567-3333> Diagnostic Sciences                      Fax:   (210) 567-3334> University of Texas                      Email: dove@uthscsa.edu> Health Science Center                    Web:   ddsdx.uthscsa.edu> San Antonio, TX    USA                   ftp:   maxrad6.uthscsa.eduthere is a program called Un-Scanit for windows which reads in pcx files (and other formats in future release) and generates a table of xy coordinate values.  The program is useful for generating xy data tables that can be used to generate plotted graphs from bitmaps.  The program is made by SILK Scientific. tel #(801) 377-6978, fax:(801)378-5474best regards,Peter CherryCity University of New YorkMedical School------------------------------Date: Fri, 28 Jun 1996 15:47:00 -0700From: "WANG, XIAOHAN" To: IT-LIST Subject: Plug-In development using visual C++Message-ID: <199606282149.OAA07605@case2.tencor.com>I'm considering writing a plug-in for Image Tool using MS Visual C++. I understand Borland C++ is recommended, but MSVC++ is what I have. This will be my first MSVC++ project. I'd like to hear any related experience/comments/advice/tips from TI authors and users.Thanks in advance -Xiaohan------------------------------Date: Sat, 29 Jun 1996 09:54:53 -0400 (EDT)From: "Dr. Norman Capra" To: it-list@sparky.uthscsa.eduSubject: Re: Specialized X,Y-Coordinate Value OutputMessage-ID: This is a generic request that seems to be related to the needs of the X-Y coordinate folks (useful for me) and the person who  asked about "trace measure".  Object classification and thresholding are tantalizing but basically of little use to me.  I need to perform a fairly large number of manual tracings to obtain areas, which the area tool provides very nicely.  The disance tool also works well to obtain min/max diameter but I can't seem to combine the measures on the same results spreadsheet.  The process also requires multiple measures of a single structure.  Object classification provides a myriad of useful measures that are not selectable for manual measurements (or I haven't figured out how). NIH Image does this nicely but I have resisted buying a Mac.  Are there plans to develop a selectable menu of area-related parameters for use with the area tool (e.g. min. diameter. max diameter, feret X,Y, etc.) has someone else done this, or is the function already available in IT but well hidden to the average user?I have really enjoyed the program and I am looking forward to its continued development. Kudos to the team at UTHSC!Thanks------------------------------End of IT-LIST Digest 26