Optical comparators and vision measurement are non-contact inspection methods used to quantify the physical dimensions, shapes, and positional relationships of manufactured parts. Unlike contact methods (e.g., coordinate measuring machines, or CMMs, that use a physical probe to touch a part), these systems rely on light and image processing to generate measurements.
The core methods covered here operate in two primary modes:
A foundational technology common to both methods is edge detection: the process of identifying and defining the boundaries of a part or its features within an image. For profile projection, edge detection identifies the outline of the silhouette; for 2D vision systems, it identifies the edges of holes, slots, or other surface features.
In plain language: these methods answer the question, “Is this part the right size and shape?” without touching it.
Precision measurement directly impacts three core manufacturing outcomes: part function, assembly compatibility, and production cost.
For components where dimensional deviation causes failure, even small errors can lead to catastrophic results. For example:
Parts must fit together consistently to enable automated assembly. If a pin’s diameter is too large, it will jam in a hole; if too small, the assembly will be loose. For high-volume production, consistent fit reduces rework and scrap.
Early detection of dimensional errors reduces waste. For example, a part machined on a Gree CNC GA-DV750 vertical machining center (positioning accuracy: 0.003 mm) can be inspected mid-production to catch tool wear before it produces 100 defective parts. Without timely inspection, scrap costs can multiply, especially for high-value materials like titanium or hardened steel.
Below is a breakdown of the practical shop-floor methods for profile projection, 2D vision systems, and edge detection, along with their accuracy limits.
Optical comparators work by passing a collimated (parallel) light beam through a part, projecting its magnified silhouette onto a screen. The silhouette is then compared to a master template or measured directly.
Profile projection accuracy depends on magnification and edge quality. Typical shop-floor systems have an accuracy of ±0.005 mm to ±0.02 mm. Higher magnification improves resolution but reduces the field of view (FOV), so operators must balance magnification and coverage.
2D vision systems use digital cameras, lenses, and lighting to capture images of parts. Software processes these images to measure features.
2D vision system accuracy is typically ±0.002 mm to ±0.01 mm for small FOVs (≤100 mm × 100 mm). Larger FOVs reduce accuracy, as pixel size increases.
Edge detection is the algorithmic process of identifying boundaries in an image. It is critical for both profile projection and 2D vision systems, as all measurements rely on correctly defined edges.
Edge detection accuracy is limited by image resolution, lighting, and edge quality. Typical limits are ±0.001 mm to ±0.005 mm for high-quality images.
Even the most precise systems produce inaccurate measurements if errors are not controlled. Below are the most common sources:
If a part is not aligned correctly in the inspection system, measurements will be skewed. For example, a shaft tilted by 1° in an optical comparator will have a measured diameter 0.017% larger than its true diameter. For a 10 mm diameter shaft, this equals a 0.0017 mm error.
Inconsistent lighting changes edge contrast, leading to incorrect edge detection. For example:
Rough surfaces, burrs, or discoloration can distort edges. For example, a burr on the edge of a machined part will make the silhouette appear larger than the true part, leading to a measured diameter error of 0.01 mm or more.
Inspection systems require regular calibration to ensure accuracy. Over time, temperature changes, mechanical wear, or software updates can cause calibration to drift. For example, a 2D vision system calibrated at 20°C may have a 0.003 mm error at 30°C due to thermal expansion of the camera or lens.
Edge detection algorithms can fail if edges are not well-defined. For example, a part with a rounded edge (rather than a sharp edge) will have a gradual brightness gradient, making it difficult for gradient-based algorithms to identify the true edge position.
Optical comparators and 2D vision systems are powerful tools, but they have critical limitations that users often overlook:
These methods only measure 2D features. They cannot detect 3D deviations such as:
For example, a part machined on a Gree CNC GA-UHD500 high-precision 5-axis machining center (positioning accuracy: 0.006 mm) may pass a 2D inspection but fail a 3D inspection if its surface has a 0.005 mm height deviation.
Profile projection only measures the cross-section of the part along the light source’s axis. If the part is not aligned correctly, the silhouette will not represent the true cross-section. For example, a hexagonal shaft tilted by 30° will have a silhouette that appears as a rectangle, leading to incorrect measurements.
Transparent parts (e.g., glass, plastic) allow light to pass through, making it difficult to create a clear silhouette for profile projection. Highly reflective parts (e.g., polished metal) can cause glare in 2D vision systems, distorting edge detection.
Larger parts require larger FOVs, which reduce measurement accuracy. For example, a 2D vision system with a 100 mm × 100 mm FOV may have an accuracy of ±0.002 mm, but a 1000 mm × 1000 mm FOV may only have an accuracy of ±0.02 mm.
Users often assume that edge detection algorithms are 100% accurate. However, edge detection is only as good as the image quality. If the image is blurry, has low contrast, or contains noise, edge detection will be inaccurate. For example, a blurry image of a hole may lead to a measured diameter error of 0.01 mm or more.
| Method | Typical Accuracy | Effort (Setup/Operation) | Typical Use Case |
|---|---|---|---|
| Profile Projection (Template Comparison) | ±0.01 mm to ±0.02 mm | Low (fast setup, visual operation) | High-volume inspection of simple parts (fasteners, stampings) |
| Profile Projection (Digital Measurement) | ±0.005 mm to ±0.01 mm | Medium (moderate setup, automated operation) | Mid-volume inspection of medium-complexity parts (shafts, brackets) |
| 2D Vision System (Fixed-Position) | ±0.002 mm to ±0.01 mm | Medium (moderate setup, automated operation) | High-volume inspection of identical parts (circuit boards, machined components) |
| 2D Vision System (Robotic) | ±0.005 mm to ±0.02 mm | High (complex setup, automated operation) | Low-to-mid volume inspection of complex parts (automotive components, turbine blades) |
| Edge Detection (Gradient-Based) | ±0.002 mm to ±0.005 mm | Low (built into most systems) | Inspection of parts with sharp, high-contrast edges |
| Edge Detection (Sub-Pixel) | ±0.001 mm to ±0.003 mm | Medium (requires high-resolution cameras and good lighting) | High-precision inspection of small parts (medical devices, micro-components) |
Use this checklist to ensure accurate, reliable inspection:
No. Optical comparators only project a 2D silhouette of the part, so they cannot measure depth or other 3D deviations. For hole depth, you would need a contact probe or a 3D vision system.
Choose profile projection for simple parts where a cross-section is sufficient (e.g., shafts, fasteners). Choose 2D vision systems for parts with surface features (e.g., holes, slots, markings) or for in-line inspection.
The maximum part size depends on the system’s FOV. Most shop-floor 2D vision systems can inspect parts up to 1000 mm × 1000 mm, but accuracy decreases as FOV increases. For larger parts, you may need to use a robotic vision system or a CMM.
Calibrate at least once per shift, or after any temperature change of more than 5°C, or after any major adjustment (e.g., lens change, fixture replacement). For high-precision applications, calibrate before each inspection run.
No. Optical inspection systems are faster and non-contact, but they cannot measure 3D features or complex geometries as accurately as CMMs. For critical 3D measurements, use a CMM. Optical inspection is best for high-volume, 2D feature inspection.
Send us your part, batch size and target port. We reply with a configuration, a binding lead window and a total-cost comparison.
Request a quote →