IPSDK Library- Main image processing features

IPSDK LIBRARY- MAIN IMAGE PROCESSING FEATURES
Reactiv’IP
IPSDK Library- Main image
processing features
26/05/2014
Copyright © 2014 Reactiv’IP, tous droits réservés. Reactiv’IP SAS, capital de 36000 €, Siret: 797 487 204,
C/O AEPI, 1 place Firmin Gautier, 38027 Grenoble Cedex 1 Tél.:+33 (0)4.76.70.97.29– Mél: [email protected]
1/4
IPSDK LIBRARY- MAIN IMAGE PROCESSING FEATURES
I. Introduction
Most of IPSDK algorithms available manage 5 Dimensions (X,Y,Z, stack of images and color
planes).
These algorithms work on all image types: 8, 16 or 32 bits and floating images. All these features
are automatically vectorized and distributable on PC cluster to ensure optimal execution
speeds.
This list is not exhaustive. It is simply the first identified features that will be available in early
versions of the product.
Functions of the IPSDK library are available in C++, Python and Java.
IPSDK is LINUX and Windows compatible.
II. Main IPSDK features
Image Edition Algorithms



Creation, conversion, cut, paste, pattern image or random image generation,
Color display conversion, color image separated in a sequence of plans, and also conversion
from Bayer to color image.
Sequence concatenation, color sequence conversion and also sequence projection (zstack).
Binarization Algorithms

Threshold, interactive threshold, Otsu method, hysteresis threshold, color threshold,
automatic threshold, labeling.
Point Operation Algorithms




Global or local histogram equalization, normalization, background correction.
Arithmetic operations, linear combinations, masks.
Absolute value, minimum or maximum between two images
Logical operations, shifting, inversion
Morphology Operation Algorithms





Erosion, dilatation, opening, closing, thickening, thinning for several kernel types in binary,
greylevel or color.
Filtering top hat, hole fill, reconstruction, edge detection, ultimate erosion.
Skeleton, centroid, pruning, triple points, terminal points, isolated points.
Morphological filters.
Watershed, distance function, Greylevel reconstruction, minimum and maximum, binary and
greylevel separation
Filtering Algorithms



FIR, several predefined filters, possibility to edit your own filters, separable filters, recursive
filters,
Adaptative filters: Bilateral, Despeckel, Nagao, Nagmod, Sigma, SNN, anisotropy
Nonlinear filters: median, delineate, deblur.
Frequency Algorithms

Fourier transformation (any image size), reference change (x,y) to (r, theta) and in the other
side, centering..

Periodic noise removing, Butterworth filter, Ring Artifact Reduction.
Copyright © 2014 Reactiv’IP, tous droits réservés. Reactiv’IP SAS, capital de 36000 €, Siret: 797 487 204,
C/O AEPI, 1 place Firmin Gautier, 38027 Grenoble Cedex 1 Tél.:+33 (0)4.76.70.97.29– Mél: [email protected]
2/4
IPSDK LIBRARY- MAIN IMAGE PROCESSING FEATURES
Differential Algorithms


Morphological gradient, linear gradient, recursive gradient, transformation (x, y) to (module,
phase).
Linear Laplacian, recursive Laplacian, Laplacian passing by 0.
Polygonal Representation Algorithms


Polygon of objects edges from Binary images, polygonal approximation.
Interior length, convex hull, circle approximation, ellipse approximation, ..
Geometrical Algorithms


Sampling, zoom, translation, rotation, symmetries
Images connection (warping), corner detection, images registration, sequence registration.
Correlation and Hough Algorithms


Any type of correlation, maxima extraction correlation.
Lines, Circles and any kind of shape Hough algorithms, maxima Hough extraction.
Segmentation algorithms



1st and 2d rank texture parameter calculation, Law and Gabor filter parameters.
Karhunen Loeve algorithm, automatic classification, interactive classification, region growing.
closure of gradients using constraints
Analysis algorithms



Several global measurements: area perimeter, inertia parameters, intercepts, Euler number,
connectivity and tortuosity.
Statistical analysis and histograms, class histograms, image comparison.
Individual analyze, individual analysis filters, sieve classification. Several individual
measurements such as area, perimeter, gravity center, inertia moment, shape factors, Ferret
diameters, diameter variations, length, width, eccentricity, orientation, difference from a
circle, from an ellipse, Greylevel , maximum, minimum, variance.
III. Ensure the stability of the product
In order to provide users an irreproachable quality of product, three levels of testing are
implemented on a daily basis for all the features marketed.
III.1. Unit tests
These tests are systematically encoded to each new development. They are used to check all
basic functions and the accuracy of results obtained from all data patterns allowed (different types of
images or numerical values).
III.2. Regression tests
These tests verify that the results obtained using the new versions of the library are exactly similar
to the results obtained using the previous version. This mechanism provides users the sustainability of
their qualification process at each change of version of the library.
Copyright © 2014 Reactiv’IP, tous droits réservés. Reactiv’IP SAS, capital de 36000 €, Siret: 797 487 204,
C/O AEPI, 1 place Firmin Gautier, 38027 Grenoble Cedex 1 Tél.:+33 (0)4.76.70.97.29– Mél: [email protected]
3/4
IPSDK LIBRARY- MAIN IMAGE PROCESSING FEATURES
III.3. Performance tests
These tests compare the execution times of all the features on equivalent machine. This control
ensures performance at least similar to each version change.
Copyright © 2014 Reactiv’IP, tous droits réservés. Reactiv’IP SAS, capital de 36000 €, Siret: 797 487 204,
C/O AEPI, 1 place Firmin Gautier, 38027 Grenoble Cedex 1 Tél.:+33 (0)4.76.70.97.29– Mél: [email protected]
4/4