Scilab at INRIA: a free alternative to other competitors (Matlab, Labview).
Version 4.1.2 (Several improvements since version 4.0, including the complete
rewrital of the "Matlab to Scilab converter" and inclusion of many standard Matlab functions)
Scilab contributions: several user-contributed toolboxes. Some related to image processing are detailed below
SWT - Scilab Wavelet toolbox: Version 4 (2007/07/08)
FracLab - Scilab Image Processing toolbox:
Version 2.03 (2006/05/11); FracLab is a general purpose signal and image processing toolbox based on fractal and multifractal methods (available for Matlab as well)
SIP - Scilab Image Processing toolbox:
Version 0.3.1 (with SIP 0.4.0 released on 13 Jul 2004); morphological operations
(erosion, dilation, watershed), edges and contours functions, filtering,
hough transform, and many more
SIVP - Scilab Image and Video Processing toolbox:
Version 0.3.0 (2006/07/15)
COSMAD -
COvariance Subspace Modal Analysis and Diagnosis (IRISA):
Version 3.0 (3.1.2 beta on 26 May 2004)
ANN -
Artificial Neural Networks Toolbox:
Version 0.4.2
Presto-Box -
Pattern REcognition Scilab ToolBOX:
Version 1.1
Featflow is an Open Source FEM package, particularly
geared for Computational Fluid Dynamics (adaptation to SciLab seems to be "work in progress")
The standard approach to represent two-dimensional data uses orthogonal
lattices (rows and columns of pixels). Quincunx or hexagonal sampling is believed to offer some advantages,
e.g. a higher degree of symmetry and a better packing density, despites its apparent implementation complexity.
Hex-Splines: the EPFL webpage on a spline family for hexagonal lattices works on image processing
on hexagonal lattice and related algorithms (for cooccurence matrix, edge
detection,...)
Image sampling grid (by David Martindale) contains images of grid patterns used in image sensors.
Lee Middleton works on image processing
on hexagonal lattice and related algorithms (for cooccurence matrix, edge
detection,...)
James Dow proposes a filter banks methodology for applications on a hexagonal
grid
LISQ is a MATLAB toolbox for the lifting scheme on quincunx grids by PM de Zeeuw [source code]
Color Filter Array (CFA) and the Bayer color coding
The Color Filter Array FAQ: delas with the application of color filter mosaic array (CFA)
for images sensors. Focused on RGB patterns and variations on the Bayer pattern
Ting Chen proposes "A Study of Spatial Color
Interpolation Algorithms for Single-Detector Digital Cameras"
Amos Storkey and possible improvements, "Renewal Strings"
algorithm
Color or channel transformations
RCT: Reversible color transform. An integer
color transformation (similar to RGB/YUV)
in "luminance/chrominance" components, targeted for lossless coding
(e.g. in JPEG 2000 with the 5/3 binary wavelet).
The notation _[x]_ denotes rounding towards the lower integer.
Forward transform
Y = _[1/4(R + 2G + B)]_
Cb = B - G
Cr = R - G
Inverse transform
G = Y - _[1/4(Cb+Cr)]_
R = Cr + G
B = Cb + G
Let It Wave (commercial site): "Let It Wave photo ID"
(link) uses bandlets (based on the work of E. Le Pennec and S. Mallat) to encode
images on as few as 500 bytes, for storage in identification cards or fast transmission (examples and comparison with JPEG 2000)
Latest MS format, alledgely close to JPEG 2000 in
quality and close to JPEG in complexity.
Features: lossless or lossy compression; with reversible color transform, for up to 16-bit integer and 32-bit floating point images, possibly n-channel data
Probably interesting and not unrelated links: Rico Malvar page for papers and patents on biorthogonal
lapped transforms, quantizations techniques and reversible color transforms. For instance, Fast progressive image coding without wavelets, presented at DCC 2000.
Wu, Ru-Shan at University of California Santa
Cruz
Geophysics
Seismic processing guides and tutorials, a collection of freely available tutorials, guides, lecture notes and primers on geophysical/seismic/seismology with signal/data processing
Séminaire Parisien d'Optimisation, seminar
organized by J.-B. Baillon, F. Bonnans, G. Carlier, P.-L. Combettes, S. Sorin
(Institut Henri Poincaré, 11 rue Pierre et Marie Curie, Paris 5e)