IPSDK has a large set of filtering and denoising features, such as anisotropic filtering. Its effect is illustrated in the figure below, where this filtering is applied on a very noisy electron microscopy image. This algorithm performs better and faster filtering than the Non-Local Mean algorithm.

Initial image

Anisotropic diffusion

IPSDK has material characterization features. This way, it is possible to determine the porosity in a given material image using denoising, binarization, split and labellization features. It is also possible to process measures on the identified elements.

Initial image

Volume with pore surimposition

Display of the pores only

IPSDK allows to process segmentation and to analyse very quickly the porosities inside a rock. The propagation features also allow to characterize the channel organisation to extract its tortuosity or the shortest path to cross the sample.

Original image

Propagation distance cartography along the porosities.

IPSDK presents some interest in the context of industrial control. Indeed, it is possible to detect material defects thanks to tools available in the library. It is also possible to characterize these defects computing many measures.

Original image (example 1)

Porosity detection, measure and visualization.

Original image (example 2)

Porosity detection, measure and visualization.

When it is impossible to correctly separate the objects in an image, the granulometry based on successive openings prove to be very useful. The main issue of this approach is its rather important calculation time. The use of extremely fast IPSDK morphological operations allows to free the process from this drawback.

Moreover, it is possible to use this algorithm on 3D and large 2D images thanks to these accelerations. Finally, the use of an exact distance map allows to apply perfectly circular structuring elements and thus improve the measure accuracy.

Initial image

Granulometry based on successive openings result

Ratio between the deleted surface and the whole grain surface depending on the kernel size

IPSDK proposes a set of functionalities specificly dedicated to the metal working industry. Specifically, the library allows to analyse polished slices to identify the various grains in the image. Once the grains segmented, this module proposes two approaches to compute the ASTM coefficient, both are described by the ASTM E112 grain size norms: the methods of intercept and planimetry.

Original image

Grain identification

Grain size histogram distribution

In some 2D or 3D images, the objects can be difficult to identify, as illustrated in the figure below. IPSDK proposes advanced segmentation tools allowing to split grains and thus provides very accurate size repartition measure, even for aggregated grains.

Initial image

Segmented nanoparticles

Equivalent diameter histogram

IPSDK allows to quickly process large data blocks (several GB) to carry out segmentations and to classify the elements in an image. The figure below presents the sand grain automatic detection and then the automatic classification into two categories according a sphericity.

Initial image

Grain segmentation

Classification according to the grain shapes

IPSDK allows to quickly process large 3D block data (several GB) to carry out segmentations and classify the elements in an image. The figure below presents splinters automatic segmentations and the automatic classification into two categories (tungsten and steel splinters).

Initial image

Material families identification

IPSDK proposes specific and fast tools to localize circular object shapes. The circles with well defined borders can be measured. The circles can also be partially hidden or overlap each other. These algorithms are based on the Hough transform applied to circles. A highly optimized implementation allows the use of this algorithm on images of several GB while guaranteeing very fast processing time.

Initial image

Image with the detected circles in superimposition

IPSDK has powerful 3D propagation features, as illustrated in the figure below. In this example, IPSDK allows to quickly segment the different cracks of an inconel pipe. In order to characterize the crack propagation inside the pipe, a seed based technique has been used. These seeds have been placed on the pipe contour, at the crack levels. The constrained propagation algorithms then allowed to propagate these seeds into the cracks. The distance computation of each voxel in the crack to the closest seed allowed to characterize the crack depth.

Image originale

Crack identification

Depth cartography inside a crack

IPSDK is a library dedicated to software development. In order to also propose powerful 3d visualization tools, IPSDK can directly be used in the Avizo software thanks to the Avizo Bridge module provided with the IPSDK licence. This association allows to take advantage of both the software conviviality and the performances offered by IPSDK. This way, it is possible to significantly reduce processing times.

Illustration of the IPSDK median filter usage in Avizo.