lp:imagep
- Get this branch:
- bzr branch lp:imagep
Recent revisions
- 45. By Tomio
-
A bunch of minor changes. Most important, now a wheel can be created using:
python setup.py bdist_wheel
and then pip installs the wheel all right. - 43. By Tomio
-
A couple of minor improvements:
The windows compiler now can compile the DLL, but outside of thes setup.
With python 3.10 they started phasing out setup tools, so the dynamic library
has to be copied manually.I have also improved the C source of the distance filter. It had issues with
cases where background area was only in the middle of the image. The current
code should behave properly for all cases. - 41. By Tomio
-
Minor changes, and two more examples to use the package. One is estimating the size of a 3D structure just summing up nonzero intensities in a volume, the other is to identify pore sizes in a 3D stack slice by slice. Both use config files to get parameters for their operation. Config values can be established reading the beginning of the script.
- 40. By Tomio
-
A couple of fine changes happend in the past few weeks, months. These include a fix for the read_img() function to load colored images as three image stacks, Improving the non-zero padding of Gaussian filters, etc. More details are in the DevelopmentNote
s.txt. - 39. By Tomio
-
Minor changes in the main code.
This version contains several example scripts for image processing, relying on this package and the BatchAnalyzer for some I/O, e.g. simple reporting and data dumping to text files.Circl-analysis is to detect circles in fluorescence images by finding the negative patches within.
Gauss-edge is to use an asymmetric, rotated Gaussian kernel to detect lines or edges in images, for improved orientation statistics. For those, who need it, the structure tensor method is also available in the ImageP package, so one can play around with these for edge detection, etc.
RadialProfileAn
alysis is a simple tool to detect point distribution by distance from the center of mass. Naturally, after some background correction, potential dynamic range compression with a power law tranformation and thresholding. It was originally meant for statistics of neuron images with explants in the middle. The current GaussKernel based filters use non-zero padding at the image edges decreasong the high intensity frames there. Naturally, this is a bit different artefact replacing a common one. The assumption is here that the intensity is not zero outside of the image, but the same as the last pixel. This works only for filters which can be run in 1D.
Display got some changes too to catch up with the development of matplotlib. It should handle now the figure objects a bit better. Allowing for overplotting information easier.
Further details are in the DevelopmentNote
s.txt file.
Branch metadata
- Branch format:
- Branch format 7
- Repository format:
- Bazaar repository format 2a (needs bzr 1.16 or later)