Difference between revisions of "User:Carandraug"

From Octave
Jump to navigation Jump to search
(→‎Bugs: follow bug #46547 - image package: canny detector in edge() should use LoG as first step)
(add list of things I would like to see on the image package soon)
 
(9 intermediate revisions by the same user not shown)
Line 3: Line 3:
 
Since then, I have contributed to Octave many times, became maintainer of the Octave Forge project and its image package. While I have no formal education in programming or computer science, it was the freedom of Octave and its community that allowed me to learn so much.
 
Since then, I have contributed to Octave many times, became maintainer of the Octave Forge project and its image package. While I have no formal education in programming or computer science, it was the freedom of Octave and its community that allowed me to learn so much.
  
= Octave to do list =
+
=== Roadmap ===
You can add stuff to my todo list (I may however move it back to your todo list on your userpage). Don't forget to sign each item with <nowiki>~~~~</nowiki>
 
  
== Bugs ==
+
The following are missing targets that would be nice to have on future releases of the image package
  
* {{Bug|43908}} - Some OF packages won't install with 64-bit indexing Octave
+
* add ellipse properties to {{codeline|regionprops}} (see bug {{bug|44100}})  ([http://hg.code.sf.net/p/octave/image/rev/7cf22a478792 done])
* {{Bug|44095}} - readline configuration - tab autocompletion of paths
+
* speed up {{codeline|regionprops}} for large images with many small objects (probably work around for loop with labeled == labelid)
* {{Bug|45088}} - image package: rangefilt requires domain and image to have equa number of dimensions
+
* support passing {{codeline|bwconncomp}} output to {{codeline|regionprops}} (and then fix {{codeline|bwpropfilt}} to use that instead of a labeled image)
* {{Bug|45333}} - image package: grayhresh fails if input doesn't have different values
+
* ND support on {{codeline|imfilter}}
* {{Bug|45367}} - addParameter and partial matching for inputParser
+
* check if {{codeline|normxcorr2}} is Matlab compatible and investigate making use of {{codeline|fftconvn}} since it's likely to receive large input
* {{Bug|45369}} - pkg build leads to duplication of PKG_ADD and PKG_DEL files
+
* investigate implementing the missing deconvolution functions (see patch {{patch|8571}})
* {{Bug|45568}} - imfilter incompatible to MATLAB for even filter width
+
* anything else that is done or submitted before the above are done
* {{Bug|45877}} - mahalanobis is wrong (and probably belongs to the statistics package)
 
* {{Bug|46011}} - Can't warn of automatic broadcasting without triggering bug #36743
 
* {{Bug|46065}} - conflicting declarations of max_align_t since gnulib update
 
* {{Bug|46099}} - image package: edge(I,'canny') gives very bad quality results
 
* {{Bug|46124}} - image package: bwfill should use 4-connectivity as default
 
* {{Bug|46160}} - image package: normxcorr2 returns complex values sometimes
 
* {{Bug|46200}} - octave automatically converts java strings into char arrays
 
* {{Bug|46392}} - image package - regionprops(bw, 'Perimeter') gives incorrect result
 
* {{Bug|46547}} - image package: canny detector in edge() should use LoG as first step
 
 
 
== Patches ==
 
 
 
* {{Patch|8713}} - image package: new color conversion functions: rgb2lab etc.
 
 
 
== octave code to write ==
 
* implement {{Codeline:inputParser.StructExpand}}
 
* make test and demo deal with tests and demos in .cc packages
 
:I don't like the first option. It's ugly and further pollutes the user namespace. And even if we have some Makefile that copies the .cc file at pkg install time, users won't know they have to make the call to test and demo differently. I guess demo and test functions would have to be modifed for that. I say just place the demos and tests on the .cc file. Later we can change pkg to copy the .cc file, and then change demo and test to look for those files if the function is .oct file. Tests are mainly used in development so users won't need them. Demos... well in the case of ellipj, it has not been working for a very long time so my guess is that people haven't been using it that much. What do you think?</pre>
 
* replacement for [http://www.mmorph.com/ morphology toolbox]
 
* port omero toolbox
 
 
 
== octave-forge to do ==
 
* finish zenity package
 
** for scale see https://savannah.gnu.org/bugs/?33330
 
* get hough transform function in image package compatible
 
* look into [[http://siptoolbox.sourceforge.net/ SIP]] and [[http://sivp.sourceforge.net/ sivp]]
 
* from Perl
 
** File::Temp
 
** check for revision numbers
 
** Getopt::Long
 
* port ReadImageJROI
 
* make compatible normxcorr2: [[http://www.idiom.com/~zilla/Papers/nvisionInterface/nip.html Jordi metioned this paper]] and
 
* fix on optim package
 
** move test scripts into function test blocks
 
** move example scripts into demo blocks
 
* default __spatial_filtering S to zeros of same dimension as DOMAIN
 
* implement im2java2d
 
 
 
== other octave related todo ==
 
* update gtksourceview octave lang file
 
* make gtksourceview highlight octave help text as texinfo
 
* make gtksourceview highlight test and demo as octave and not comment
 

Latest revision as of 21:50, 31 August 2018

I got involved with Octave on the Winter of 2010, at the end of my first year as PhD student. My project involved using a Matlab piece of code written by a collaborator but I refused to use it, I wanted free software. So I convinced the collaborators to release it under GPL and ported it to Octave (released as frapinator, but I really should go back and write it properly).

Since then, I have contributed to Octave many times, became maintainer of the Octave Forge project and its image package. While I have no formal education in programming or computer science, it was the freedom of Octave and its community that allowed me to learn so much.

Roadmap[edit]

The following are missing targets that would be nice to have on future releases of the image package

  • add ellipse properties to regionprops (see bug #44100) (done)
  • speed up regionprops for large images with many small objects (probably work around for loop with labeled == labelid)
  • support passing bwconncomp output to regionprops (and then fix bwpropfilt to use that instead of a labeled image)
  • ND support on imfilter
  • check if normxcorr2 is Matlab compatible and investigate making use of fftconvn since it's likely to receive large input
  • investigate implementing the missing deconvolution functions (see patch #8571)
  • anything else that is done or submitted before the above are done