Image Analysis for Academia and Research

The discipline the platform was built for, and the one where standards are strictest. Chart comparison and hand-placed measurements give way to standard-referenced numbers that transfer between operators, cohorts and institutions.

• Grain size, Jeffries & intercept ASTM E112 / E1382
• Phase & area fraction ASTM E562 / E1245
• Inclusion rating by type and severity ASTM E45 / ISO 4967
• Banding & microstructural orientation ASTM E1268
• Nodularity and graphite classification ASTM A247 / ISO 945
• Microhardness & case depth profiling ASTM E384 / E92
• Dendrite arm spacing & solidification
• Decarburization and surface layer depth

Device reliability research lives in cross-sections: what is the layer thickness, how much void is in the joint, did the intermetallic grow after thermal cycling. These are counting and thickness problems at small scale, on images that are often SEM rather than optical.

• Solder joint void fraction and distribution
• Intermetallic compound layer growth after aging
• Plating and metallization thickness ASTM B 487
• Via, through-hole and trace geometry measurement
• Die attach and underfill void quantification
• Conformal coating uniformity and coverage
• Particulate contamination on assemblies ISO 4406 / 4407

Optical components fail on defects that are too small and too numerous to count by eye, and inspection by a trained technician is famously inconsistent between people. Automated counting turns a subjective surface assessment into a defect density with a size distribution behind it.

• Surface defect, scratch and dig counting by size class
• Bubble and inclusion counting in glass and crystals
• Fiber optic cross-section: core, cladding, concentricity
• Polishing and lapping damage assessment
• Edge chip and subsurface damage quantification
• Waveguide and micro-optic geometry measurement

One of the fastest-growing areas in academic characterization, and one where phase contrast defeats classical methods. Trainable detection has been applied to demanding cases such as cross-sectioned lithium-ion cathodes.

• Cathode and anode cross-section particle detection
• Electrode coating thickness and uniformity
• Electrode porosity and active material fraction
• Binder and conductive additive distribution
• Membrane and separator layer measurement
• Cracking and degradation after cycling

In a process-parameter study the defect population is the dependent variable, which means the analysis has to be byte-for-byte identical across every build in the matrix. Locked routines make a printer parameter sweep comparable months apart.

• Porosity fraction, size and spatial distribution
• Lack-of-fusion versus spherical gas pore classification
• Melt pool geometry and overlap measurement
• Feedstock powder sizing, shape and satellites
• Build-direction grain morphology and anisotropy
• Post-processing effect on defect population

Composite and ceramic performance is governed by what the microstructure contains and how it is distributed. These measurements are the ones that correlate with the mechanical test the group already runs.

• Fiber volume fraction in laminates
• Fiber length, diameter and orientation distribution
• Void content and delamination in composites
• Filler and reinforcement dispersion quality
• Ceramic grain size and sintering evaluation
• Ceramic and refractory porosity quantification
• Spherulite and polymer phase morphology

Thin section petrography is point counting by another name, and it is exactly the task automation was invented for. Color and texture-based detection separates phases that a threshold cannot.

• Mineral grain size and shape distribution
• Modal composition by phase area percentage
• Porosity and pore connectivity in thin section
• Mineral liberation and particle counting in ore
• Coal petrography and maceral quantification
• Fracture and vein network characterization

Formulation and product structure research needs number-based distributions with the images retained, not just an ensemble average. Every measured object stays inspectable when a result needs defending.

• Active ingredient particle sizing and shape
• Tablet and capsule coating thickness
• Excipient dispersion and agglomeration
• Food microstructure and color distribution
• Starch granule and fiber measurement
• Foreign particle detection and classification

The core of quantitative metallography and the measurements most theses open with. AI-assisted boundary detection copes with twins, incomplete boundaries and the uneven etch quality that is routine when samples are prepared by students.

• Grain size, Jeffries & intercept ASTM E112 / E1382
• Grain boundary detection with twin tolerance
• Phase & area fraction ASTM E562 / E1245
• Duplex and multi-phase separation on similar grey levels
• Banding & microstructural orientation ASTM E1268
• Cell size in anodized aluminum
• Alpha / beta constituent quantification in titanium

Everything the material should not contain, sized, counted and classified by morphology. In process-parameter studies these are usually the dependent variable, so the analysis has to be identical across every specimen in the matrix.

• Porosity, void fraction and size distribution
• Lack-of-fusion versus spherical gas pore classification
• Inclusion rating by type and severity ASTM E45 / ISO 4967
• Micro-crack and discontinuity quantification
• Particle counting on cleanliness filters ISO 4406 / 4407
• Weld and heat-affected-zone defect measurement

Per-object measurement rather than an ensemble average, so every outlier keeps the image behind it. When a reviewer asks what the tail of the distribution actually looks like, the answer is a micrograph, not an assumption.

• Particle sizing, counting and distribution
• Shape descriptors: circularity, aspect ratio, convexity
• Satellite and agglomerate detection
• Additive manufacturing feedstock characterization
• Controlled powder dispersion for preparation
• Fiber length, diameter and orientation in composites

Thousands of perpendicular measurements along a cross-section instead of a handful of hand-placed calipers. Thickness works on any detected shape including open contours, which extends the method well past anything designed as a coating.

• Coating thickness with full statistics ASTM B 487
• Multilayer stack measurement, layer by layer
• Case depth and diffusion layer profiling
• Thermal spray porosity, oxides and unmelted particles
• Electrode and thin-film coating uniformity
• Thickness on open contours: walls, membranes, interfaces

The measurements that link processing conditions to structure. Automating them removes the operator-to-operator scatter that otherwise competes with the effect being studied, which matters when the cooling-rate correlation is the finding.

• Secondary dendrite arm spacing
• Nodularity and graphite form classification ASTM A247 / ISO 945
• Nodule count per unit area
• Eutectic fraction and morphology
• Shrinkage versus gas porosity discrimination
• Columnar to equiaxed transition mapping

Automatic indent detection, diagonal measurement and conversion, with unattended traverses and grids. An afternoon at the eyepiece per specimen becomes an overnight run and a morning of data.

• Vickers & Knoop automatic indent measurement ASTM E384 / E92
• Case depth traverses with automatic profiling
• Hardness maps across welds and heat-affected zones
• Indent validation, flagging and automatic retest
• Correlation of hardness with local microstructure