Glossary · Machine vision
Edge detection
Also known as: Edge finding
German: Kantenerkennung
In image processing, edge detection locates positions where image brightness changes sharply, which usually correspond to object boundaries; the results are used for measurement, alignment and feature extraction.
- Machine vision
In one sentence
Edge detection finds sharp brightness changes in an image, usually object boundaries, for measuring, aligning and extracting features.
Example
Edge detection along two search lines measures the gap between door and fender to a tenth of a millimeter.
How it applies
- Engineering: Edge tools can find edges with subpixel accuracy along a search line or region. Good contrast, low noise and suitable lighting (often Backlighting) determine accuracy. Common algorithms compute brightness gradients, for example with Sobel or Canny operators, and then locate their maxima.
- Commissioning: Edge polarity, threshold and search region are set with sample parts; dirt, burrs or reflections cause false edges.
- Documentation: The documentation team should document the edge tools, search regions and measurement limits, and how the measurement was checked against a reference (for example a gauge).
Edge detection vs. blob analysis
Edge detection finds boundaries; Blob analysis evaluates regions. Measurement tasks usually rely on edges; counting and presence tasks on blobs.