Re: morphological image processing in Python
On Mon, 3 Oct 2005 14:23:11 + (UTC), Alexandre Fayolle <[EMAIL PROTECTED]> wrote: >Hello, > >I'm about to start a project which will involve some greyscale image >processing using morphological operators (erosion, dilation, distance >transforms...), and I was wondering if these operators were available in >Python through some open source libraries. > >A quick google search returned pymorphpro[1], which is unfortunately not >free software, and there also seem to be something available in ITK [2]. >Unless I'm mistaken, the PIL does not provide these operations, nore do >Numeric/numarray/scipy. > >Have I missed the one true Python mathematical morphology toolbox? > >Thanks in advance for any advice. Actually, numarray does support morphological array operations. See the chapter on 'Multi-dimensional image processing' in the numarray docs. Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Re: Processing an image with numarray.nd image
On Tue, 04 Oct 2005 11:56:51 +0200, Raphaël MARC
<[EMAIL PROTECTED]> wrote:
>Hello,
>
>Can anyone tell me how to open an image
>and transform it into a list so that
>the functions of the multi dimensionnal module of
>numarray (numarray.nd image) can process it ?
>
>Do I have to use PIL ?
>
>So I would code something like :
>import Image
>im = Image.open("Python.jpg")
>data = list(im.getdata())
>import numarray.nd image as ti
>ti.median filter(data,...)
I've been doing something like that...
from numarray import *
import numarray.nd_image as Filter
import PIL.Image as Image
im = Image.open(file_name)
array_dat = reshape(array(list(im.getdata())), im.size)
filt_array_dat = Filter.correlate(array_dat, my_filter)
et cetera...
Matt Feinstein
--
There is no virtue in believing something that can be proved to be true.
--
http://mail.python.org/mailman/listinfo/python-list
Re: Python Newbie
On 5 Oct 2005 08:38:44 -0700, [EMAIL PROTECTED] wrote: >Hi, > >I'm learning Python now and I'm really pleased with everything so far, >however, at some point I'll want to develop a Windows application (no >other OS at this point). The question are: >1. What toolkit is best for Windows development? >2. Which toolkits can I exclude from consideration? >3. What toolkit is most popular for Windows development? >4. Are there any impartial reviews of these packages anywhere? You have to give us a clue about what you mean by 'best'. Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Re: Learning Python
On Mon, 10 Oct 2005 12:50:18 -0400, Paul DiRezze <[EMAIL PROTECTED]> wrote: >I'm spending the next two weeks off and I'm looking to take a crack at >learning how to program in Python. Here's a list of the places I've >bookmarked: > >http://www.python.org/doc/ and more specifically >http://wiki.python.org/moin/ >http://wiki.python.org/moin/BeginnersGuide >http://www.python.org/doc/Intros.html >http://www.python.org/topics/ > >I'm looking for additional resources (links, names of books, >whatever...) that you think may help me out. > >I'm a hobby programmer using mostly BASIC(s) and some Java. I know >procedural programming and I know what encapsulation, inheritance and >polymorphism are but I have very little experience in using them as I've >written just a few Java programs (i.e., I guess I could use a primer in >OO programming). > >Any ideas? http://diveintopython.org/ Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Re: Installing Python at Work
On 17 Oct 2005 08:12:35 -0700, "Nikola" <[EMAIL PROTECTED]> wrote: >I'm currently learning Python for my own use. >I'm considering installing it on a work laptop, knowing that it is >non-licensed, distributable software. > >However, does it access communication ports? I know the company checks >their ports regularly for activity. > >I won't be doing anything very serious; I'm just trying out Python, >learning the basics from 'Learning Python' by O'Reilly. Python installations are pretty benign. Note, however, that when you install additional modules, you are trusting the module writer. For locality, you can do a completely local Python installation in Linux, and generally speaking, additional modules can be installed locally as well. Under win32, the standard binary installer makes a registry entry which is checked when additional (binary) modules are installed. I don't think there's any other 'non-local' behavior. Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Re: any python module to calculate sin, cos, arctan?
On Tue, 08 Nov 2005 12:30:35 GMT, "Raymond L. Buvel" <[EMAIL PROTECTED]> wrote: >Shi Mu wrote: >> any python module to calculate sin, cos, arctan? > >The other answers in this thread point you to the standard modules. If >you need arbitrary precision floating point versions of these functions >check out: > >http://calcrpnpy.sourceforge.net/clnumManual.html Unless you're using Windows. Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Re: any python module to calculate sin, cos, arctan?
On Tue, 08 Nov 2005 06:43:51 -0800, Robert Kern <[EMAIL PROTECTED]> wrote: >Matt Feinstein wrote: >> On Tue, 08 Nov 2005 12:30:35 GMT, "Raymond L. Buvel" >> <[EMAIL PROTECTED]> wrote: > >>>http://calcrpnpy.sourceforge.net/clnumManual.html >> >> Unless you're using Windows. > >Why? Have you tried compiling it and failed? Copied from the linked site: "Windows is not supported but it may be possible to get this module to work. You will need to use the GNU tools and figure out how to build the CLN library. Then you need to figure out how to compile and link the clnum extension module for the standard Python. If anyone gets this to work and wants to contribute the results, contact me at rlbuvel at gmail dot com." Note: "may be possible", "you need to figure out", "if anyone gets this to work". The message seems clear to me-- 'If you want this capability, you can either-- 1) install all the development tools, 2) compile all the relevant libraries, and then 3) link and compile the Python extension module at your own risk and on your own (or on your company's) time-- or, use Linux. Personally, I'd use Linux. On the other hand, not everyone has that option, & it seems reasonable to me to point out that there's a non-trivial difficulty here. Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Re: Which kid's beginners programming - Python or Forth?
On 27 Jun 2005 20:16:12 -0700, "BORT" <[EMAIL PROTECTED]> wrote: >Please forgive me if this is TOO newbie-ish. > >I am toying with the idea of teaching my ten year old a little about >programming. I started my search with something like "best FREE >programming language for kids." After MUCH clicking and high-level >scanning, I am looking at Python and Forth. Both have advocates that >say each is a great approach to learning computers. FORTH is 'way outside the mainstream of current programming, while Python is, if anything, excessively buzz-word compliant. If you want to teach your kid something that will a basis for learning anything about current practices in programming, teach him Python. Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Gotchas in user-space matplotlib install?
Hi all-- I'm planning to try to do a completely local install of matplotlib (in Fedora Core 1)-- the system administrator isn't going to stop me-- but he isn't going to cooperate either. I've got the tarballs for python, numeric, numarray, matplotlib, ipython, wxpython and freetype-- which I think covers the various pre-requisites and post-requisites. One semi-obvious question is where to put the freetype library (the system version in FC1 is not up to the required level)-- but I can only wonder what other trouble I'm going to get into. Any advice before I take the plunge would be appreciated. TIA... Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Re: Gotchas in user-space matplotlib install?
>If you do a --prefix=~/usr for all 'configure' and 'python setup.py >install' commands, and set your PATH, C_INCLUDE_PATH, >CPLUS_INCLUDE_PATH, LIBRARY_PATH, and LD_LIBRARY_PATH variables >accordingly, and set the MATPLOTLIBDATA environment variable to point >to ~/usr/share/matplotlib, it should work. > >When debugging your setup, you might want to run your test >matplotlib/pylab script with > > > python myscript.py --verbose-helpful > >OR > > > python myscript.py --verbose-debug > >to get extra information about where matplotlib is looking for things. For the record, I persuaded the sysadmin to install current levels of the tcl, tcl-devel, tk, and tk-devel libraries-- my argument was that you couldn't get a useful installation of Python without them. Since I used /usr/people/mrf/python24 for my home Python directory, there were a few tweaks needed to get the matplotlib installer to find stuff. To get access to the freetype and Numeric libraries and what-have-you, adding /usr/people/mrf/python24/lib/python2.4 and /usr/people/mrf/Numeric-23.8 to the basedir dictionary in setext.py seemed to work. In addition, I think I had to copy the Numeric include directory into an additional location. There was also a little fiddling with PYTHONPATH to insure that the installer found both numarray and Numeric. Didn't seem to need MATPLOTLIBDATA, but that would have been my next try. With a working version of Tcl/Tk, the need for other widget sets became somewhat academic. For GTK, Fedora Core 1 is apparently so far behind current levels of GTK that it is pointless to complain. Wx is still a possibility. All in all, not actually excruciating-- and now I have a working version of matplotlib! Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Re: Gotchas in user-space matplotlib install?
On Mon, 15 Aug 2005 09:02:57 -0500, John Hunter <[EMAIL PROTECTED]> wrote: >>>>>> "Matt" == Matt Feinstein <[EMAIL PROTECTED]> writes: > >Matt> All in all, not actually excruciating-- and now I have a >Matt> working version of matplotlib! Matt Feinstein > >Great! While this is all fresh in your mind, would you be able to add >a wiki entry at >http://www.scipy.org/wikis/topical_software/MatplotlibCookbook > >Thanks, I'm making notes on it now. The only thing that the installation really stumbled over was some includes-- the Python installation left me with a '/usr/people/mrf/python24/include' directory, which at the matplotlib installation had 'python' and 'python2.4' subdirectories-- numarray put its includes into the 'python2.4' subdirectory, while Numeric put its includes into the 'python' subdirectory. The matplotlib installer found the numarray includes but not the Numeric includes-- copying the Numeric includes into the 'python2.4' subdirectory got the ball rolling again. Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Off-screen rendering in PyOpenGL?
Poking around in the PyOpenGL tarball... I can see that the wrapper for the WGL pixel format function includes flags for rendering to a bitmap and for hardware acceleration... so maybe I could get hardware-accelerated off-screen rendering under win32.. but what about linux? Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Matplotlib question-- Numeric or numarray?
I'm working my way through the matplotlib documentation & there's a point that's ambiguous-- the pdf file (dated Mar. 1, 2005) warns of dire consequences if I use the 'wrong' array package-- e.g., put numarray in the .matlabrc file if the compile-time package is Numeric. But there's only one current, unlabeled, windows installer and there seems to have been a change, some time back before version 0.7, in how this question is dealt with. Can someone clarify? thnksndvnc Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
My stupid newbie mistake
I named a file 'try.py' and then was stumped, for a while, when >>> import try gave a syntax error... So, how about a 'YouAreUsingAReservedWordStupid' exception ? Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Re: goto statement
On Wed, 20 Apr 2005 10:23:58 +0100 (BST), praba kar <[EMAIL PROTECTED]> wrote: >Dear All, > > In Python what is equivalent to goto statement I'd like to that implemented in an interpreted language. Requires some time travel. Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
New versions of numarray?
Hi-- I notice that there are some new versions of numarray available for download-- is there any documentation on what's new/fixed/broken? Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Re: Which IDE is recommended?
On Wed, 27 Apr 2005 21:16:29 +0800, "monkey" <[EMAIL PROTECTED]> wrote: >Read through python site for programming tool, really plenty of choices :-) >(For c++, I just can't breath with very very limited choices) > >Tried Spe, it come with wxGlade built-in very nice(is Spe still actively >develop?). But seem that Boa Constructor and PyDev(the plug-in for Eclipse) >also worth looking. Actually which one are you guys using? and why? I think >it is also valuable for those who are new to python as me. If you intend to use Python for Matlab-like calculations with numerical arrays and plotting with Matplotlib, then ipython is the right choice-- it has a special 'pylab' mode that is Matplotlib-aware, allowing you to make plots interactively (as well as various other useful features). Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Numarray question
If I try >>> 2 < array([1,2,3]) I get: array([0, 0, 1], type=Bool) which is pretty slick, However if I set >>> q = 2 < array([1,2,3]) >>> q and q I get a runtime error: "An array doesn't make sense as a truth value." So.. why not? It seems to me that if I could vectorize logical expressions (a la Matlab), it would be a lot easier to write functions that will take either scalar or vector input values. Am I missing something? Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Re: OSx 10.4 lacks pythonIDE?
On Fri, 13 May 2005 03:21:40 +0100, baza <[EMAIL PROTECTED]> wrote: >Where is the IDE in 'Tiger' for the mac? Don't tell me I have to use >text edit all the time?? You can use Spotlight to find the file idle.pyw and use that as an IDE... Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Applying a function to a 2-D numarray
Is there an optimal way to apply a function to the elements of a two-d array? What I'd like to do is define some function: def plone(x): return x+1 and then apply it elementwise to a 2-D numarray. I intend to treat the function as a variable, so ufuncs are probably not appropriate-- I realize that what I'm looking for won't be terrifically efficient, but I'd like to avoid doing it in the -worst- possible way. Some things I've looked at include things like def applyfun(m,f): elist = [f(e) for e in m] return reshape(elist,m.shape) however, I can see that this looks neat but probably generates several copies of the array, which is not so neat. Is there a better way? Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Re: Applying a function to a 2-D numarray
On Mon, 16 May 2005 11:07:06 -0600, Steven Bethard <[EMAIL PROTECTED]> wrote: > >I must be missing something, because the simplest possible thing seems >to work for me: > >py> import numarray as na >py> def plus1(arr): >... return arr + 1 >... >py> def apply_func(arr, f): >... return f(arr) >... >py> a = na.arange(20, shape=(4, 5)) >py> a >array([[ 0, 1, 2, 3, 4], >[ 5, 6, 7, 8, 9], >[10, 11, 12, 13, 14], >[15, 16, 17, 18, 19]]) >py> apply_func(a, plus1) >array([[ 1, 2, 3, 4, 5], >[ 6, 7, 8, 9, 10], >[11, 12, 13, 14, 15], >[16, 17, 18, 19, 20]]) > >Is this not what you wanted? The problem is that I chose an example function that's too simple. Non-trivial functions aren't so polymorphic, unfortunately. Sorry for the confusion. Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Re: Applying a function to a 2-D numarray
On Mon, 16 May 2005 12:03:24 -0600, Steven Bethard <[EMAIL PROTECTED]> wrote: >Can you give an example of what you really want to do? Probably there >are numarray functions that you can use. In general, you'll do better >applying a sequence of numarray functions than operating element-wise on >an array and converting it from a list back to an array... Well, for example, suppose I want to modify the elements of the matrix in some fashion. However, I'm not entirely sure how I want to do it. As a strawman, I generate a function with a Boolean test in it that multiplies by one factor if the matrix element is in an interval and by a different factor if it is outside the interval def genfunc(xmin, xmax, f_in, f_out): def booltest(x): in_interval = x > xmin and x < xmax if in_interval: return x*f_in else: return x*f_out return booltest Generating the function in this way gives me great flexibility in deciding exactly what function I apply to the matrix. It's why I want to use Python for this analysis. The part of the function I vary and play around with is localized in one place in the 'genfunc' function-- I can change that and everything else stays the same. However, I realize that the gain in flexibility means a loss in efficiency. I'm limited to not-so-efficient ways of. For this work, it's OK-- I just want to know the best not-so-efficient way of doing the calculation. Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Unhappy with numarray docs
I spent all day yesterday trying to figure out how to do file IO in the numarray module-- I -did- (I think) figure it all out, eventually, but it's left me in a rather sour mood. 1. The basic functions and methods: fromfile, fromstring, tofile, and tostring, are buried, in non-alphabetical order, in two chapters that list -all- the functions and methods. This is not user-friendly. The explanations themselves, once I found them, are OK, but a line or two of sample code is always nice. File IO is a real-world necessity in an environment where most people use Matlab and/or IDL. And is it obvious, btw, that reading data should be a function? What if I want to read into a buffer? 2. The memmap documentation is screwed up in a more serious fashion. I was able to deduce relatively rapidly that from numarray.memmap import * was needed for some of the example code. But I didn't immediately have the degree of clairvoyance needed to figure out that import numarray.numarryall as num was required for the critical step of associating an array with a MemmapSlice. Grr. Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Re: Unhappy with numarray docs
On Wed, 01 Jun 2005 08:11:36 -0700, Scott David Daniels <[EMAIL PROTECTED]> wrote: >Propose some fixes to the documents that will make this easier for >the next one in line. You don't even need to get it exactly right; >the person after you can fix the mistakes you make. This is the >process we use for this. See this as an opportunity to contribute, >not simply a frustration about how much you overpaid for the product. Which is why I was specific about what I didn't like about the documentation and why I didn't like it. Seriously, what more should I do? It's plainly inappropriate for me to write documentation for a module that I'm still struggling to learn. Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Re: Unhappy with numarray docs
On Wed, 01 Jun 2005 09:55:14 -0600, Fernando Perez <[EMAIL PROTECTED]> wrote: >Just a suggestion: post your message on the numeric discussion list (where >numarray is also discussed). Most of the num* developers only check c.l.py on >occasion, so it's very easy that your message will be simply missed by the >appropriate people, which would be a shame. They are generally very >responsive to user requests and constructive criticism. Done. Thanks for the suggestion. Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
PIL and GeoTIFF
Hi all-- I've succeeded in using the PIL (Python Imaging Library) to read a simple GeoTIFF file and to extract data from the file's GeoTIFF key-- but I'd also like to write GeoTIFFs, and there doesn't appear to be a one-step way of doing that. I suspect, first, that -writing- a GeoTIFF file with PIL means poking around in the PIL TIFF plugin file which I'm a little wary of doing. Also, I don't see any clear description in the PIL documentation on how to add specified TIFF keys to a saved file. Any words of wisdom about either of these questions would be much appreciated. Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Re: PIL and GeoTIFF
On Thu, 09 Jun 2005 14:19:21 -0400, Khalid Zuberi <[EMAIL PROTECTED]> wrote: >If you are willing to go in a bit of a different direction, GDAL >supports GeoTIFF and includes python bindings: > >http://www.remotesensing.org/gdal/ Looks interesting, but kinda heavy-duty for my purposes. If a couple of lines of Python added to what I've already got will do the job, I'd rather do that then compile and learn yet another API. Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
Problem with 'struct' module
Using the 'struct' module (Win32, python version 2.4.1)--
The library documentation says that 'no alignment is required for any
type'. However, struct.calcsize('fd') gives 16 while
struct.calcsize('df') gives 12, implying that double precision data
has to start on a double-word boundary.
Matt Feinstein
--
There is no virtue in believing something that can be proved to be true.
--
http://mail.python.org/mailman/listinfo/python-list
Re: Problem with 'struct' module
On Tue, 14 Jun 2005 14:24:56 -, Grant Edwards <[EMAIL PROTECTED]> wrote: > >Your example is not using standard alignment. It's using >native alignment: > >By default, C numbers are represented in the machine's native >format and byte order, and properly aligned by skipping pad >bytes if necessary (according to the rules used by the C >compiler). > >Alternatively, the first character of the format string can >be used to indicate the byte order, size and alignment of >the packed data, according to the following table: > > CharacterByte orderSize and alignment > @native native > =native standard > <little-endianstandard > >big-endian standard > !network (= big-endian) standard > >If the first character is not one of these, "@" is assumed. > >Native byte order is big-endian or little-endian, depending >on the host system. For example, Motorola and Sun >processors are big-endian; Intel and DEC processors are >little-endian. > >Native size and alignment are determined using the C compiler's >sizeof expression. This is always combined with native byte >order. > >Standard size and alignment are as follows: no alignment is >required for any type (so you have to use pad bytes); short is >2 bytes; int and long are 4 bytes; long long (__int64 on >Windows) is 8 bytes; float and double are 32-bit and 64-bit >IEEE floating point numbers, respectively. > >Note the difference between "@" and "=": both use native >byte order, but the size and alignment of the latter is >standardized. Thanks. I clearly missed the point of the explanation... Matt Feinstein -- There is no virtue in believing something that can be proved to be true. -- http://mail.python.org/mailman/listinfo/python-list
