Overview of Neo/NST Vision Features

Christian Lange, 2002-08-07

There are several collections of functions:

Array based tools and filters
Explorer Units (for stereo robot vision)
2d Fast Fourier Transform
Class CImage

Note: Some links to manpages on this page refer to static copies of them. So you need not to check out the source of contributed folders to read their manpages. But if you decide to use a contributed folder, you should read the check out version of the manpages because these may be more up to date.

Array based tools and filters

The units in this section use red pins to transmit images. Thus a circuit has to be remaked for processing images of varying size. The images are stored line-by-line, starting at the lower left corner. To process rgb-images three separate pins are used. Have a look at Example#7 from Section 7 (plotting of data) or Example#2 from Section 15 (Interactive programs) in the NEO-tutorial.

Image input/output

scanf: Read pgm or ppm files (folder TEXT: string)
read_image: Read pbm,pgm or ppm files (folder TEXT: file)
grab_wdw: Grab image from Window (folder RENDER: pixel)
v4l_grab: Get image from TV-card (implemented in the contributed folder v4l)
write_pix: Display image in window (folder RENDER: pixel)

Image processing

Filters: gauss, laplace, ... (see folder filter)
Segmentation: diff_segment
Applying gabor masks: pic_to_gabor
Histograms: histogram (folder RENDER: pixel)

Explorer units (for stereo robot vision)

The explorer units are implemented in contributed the folder explr and documented in the documentation of the GRAVIS circuit and their manpages:

abskor,	cone,	cort_lay,	dfadeout,
eye_filter,	eye_filter_gap,	eye_mask_gen,	eye_rec_field,
eye_rec_fields,	gazeblock,	linhisto,	moddiff,
selthr,	skin,	stereo,	stscan,
visskin.

Class CImage

The vision library neovision.so provides the class CImage for use in prog_units. Additionally there are many circuit-folders using the library.

To use CImage you need to check out the sources:
cd $HOME/mynst/
cvs co folder_vision

and compile them:
cd $HOME/mynst/
make loadlibs

Furthermore you may want to check out the folders:
cd $HOME/neo/
cvs co cv_folders

To use the csc-segmenter you need its sources:
cd $HOME/mynst/
cvs co nst_contrib/csc_nchannelsrc
mv nst_contrib/csc_nchannelsrc .

and compile them:
make loadlibs

To be able to use the vision library and the folders from Neo you need to include them into your .neofolders:

FOLDERGROUP VISION
      neovision       /homes/clange/mynst/o.linux/neovision.so
      cv_aux          /homes/clange/neo/cv_folders/cv_aux
      cv_feature      /homes/clange/neo/cv_folders/cv_feature
      cv_filter       /homes/clange/neo/cv_folders/cv_filter
      cv_io           /homes/clange/neo/cv_folders/cv_io
      cv_segment      /homes/clange/neo/cv_folders/cv_segment
      csc_nch         /homes/clange/mynst/o.linux/csc_nchannel.so