Revolutionizing Routine Metallography
Exploring Cutting-Edge Microscopy Solutions for Metal Analysis
The properties of a metal, and its behavior in service, are dominated by its microstructure. All of the key mechanical properties, the corrosion behavior and fatigue performance are governed by a combination of grain size, composition, phase size/distribution, inclusions, and localized microstructural variations.
ZEISS offers a leading-edge portfolio of solutions covering a wide range of microscopy techniques for routine metallography and quality control:
Enhanced Image Analysis Using Machine Learning
Quantitative analysis of an image or 3D data set always requires a segmentation step. Segmentation is the operation of dividing the image into multiple regions and differentiating them from each other in some way. Once this is done, the regions can be analyzed to obtain real actionable data - regions could be individual grains, inclusions, pores, different phases, layers or anything that needs examination.
Segmentation may be challenging for multiple regions
- Regions may have similar color/contrast
- Regions may only differ by texture
- Artefacts/scratches may give false positives
- Noise in 3D data sets hinders segmentation accuracy
- Analysis of multichannel (RGB or more) images is increasingly complex
ZEISS ZEN Intellesis uses guided machine learning to overcome these segmentation issues. Using training input provided by the user in a simple graphic workflow, it uses over 30 different parameters to assess each pixel and assign each the correct category. The process is repeated, training the machine learning model and improving its power and accuracy. The user can then automatically apply their machine learning to their entire data set, segmenting hundreds of images or 3D data sets into a format that is easy to analyze. With no prior expertise in machine learning needed, ZEISS ZEN Intellesis brings the power of artificial intelligence to the every day microscope user.