Typical Image Characteristics¶
← Back to Electrochemical Imaging
Images from electrochemical imaging workflows — whether from microscopy, spectroscopy, or tomography — share several features that distinguish them from natural photographs:
- Grayscale
Most SEM, TEM, and AFM images are single-channel grayscale, though pseudo-colour is often applied for visualization.
- High dynamic range
16-bit acquisition is common; BSE images may span a small fraction of the available range due to a narrow compositional spread.
- Noise
Shot noise (Poisson): dominant at low beam currents / short dwell times in SEM.
Gaussian read noise: detector electronics.
Streak / scan noise: mechanical vibration or electromagnetic interference during line scanning.
- Artifacts
Charging: bright halos and bright spots in non-conductive regions.
Curtaining: parallel vertical streaks in FIB cross-sections from differential milling rates.
Drift: image shear from thermal or mechanical drift during long acquisitions.
Beam damage: progressive structural changes under the electron beam.
- Scale and magnification variation
A dataset may contain images acquired at different magnifications, requiring normalization of the physical scale before training ML models.
- Class imbalance
In electrode cross-sections, one phase (e.g., carbon black binder) may occupy only a few percent of the image area, creating severe class imbalance for segmentation models.
- Temporal dependence
Many acquisitions are not single snapshots but ordered frame sequences or video — in-situ/operando monitoring, particle tracking, or degradation studies over time. Analysis must then account for frame-to-frame correspondence and trends across time, not just the spatial content of one image; see I/O and Tracking for how JuSPICE represents and tracks such multi-frame and video datasets.