Peak Metrology utilizes smart cameras and vision systems to quickly identify areas of interest for further detailed analysis. The vision system quickly scans the surface using a 2D camera and a 3D microscope further analyzes the individual points of interest that are found by the machine vision system. These areas of interest could be any feature or defect that is normally or abnormally occurring on a surface. This enables a “seek and destroy” method of programming inspection locations for the microscope whenever the locations are not known ahead of time.
What Problem Does This Solve?
There are many uses for high resolution 3D profilometers and microscopes. Their use is typically limited to a laboratory setting where operators are manually using the equipment. Moving from location to location on the part’s surface manually, or adjusting to location variation via joysticks. With Vision PreInspection, the user is able to pre-program a recipe that replaces the manual operator involvement in creating the coordinates that the inspection needs to take place.
Furthermore, there are many applications and uses for a 3D profilometer where the operator simply cannot navigate the part surface on their own. In these cases, the use of a separate Automated Optical Inspection (AOI) system is common. However, the coordinates of the inspection locations for the 3D profilometer still need to be translated to the profilometers coordinate frame. By combining the vision + profilometry we remove the need to transfer coordinate systems between tools. We also remove the need for the upstream AOI altogether which reduces cost and floorspace.
From Sight to Insight. All-in-one Machine Vision and Microscopy
- Macro-to-Micro Fusion: Integrates fast 2D screening with precision 3D profilometry.
- Intelligent Targeting: Uses machine vision to automatically identify defects or features for detailed inspection.
- Maximum Throughput: Triggers lengthier 3D imaging only at detected sites.
- On-the-Fly Auto-Stitching: Automatically maps stitched multi-tile 3D scans for larger defects.
- Complete 2D/3D Traceability: Pairs macro 2D visuals with precision 3D data.
Showing a vision system mounted next to the 3D profilometer.
Main Benefits of PreVision
- Faster inspection times. Skip defect-free areas and only direct the 3D microscope to areas that need inspected.
- Reduce CapEx and equipment footprint. Combine the function of 2 systems into one.
- Zero code and no operator involvement in setting up measurement locations. Replace rigid coordinate tables and manual joystick operation with dynamic vision targeting that adapts to part to part variance.
How It Works
Peak Metrology brings together the best in vision systems and 3D profilometry into a single combined tool.
Selecting the Appropriate PreVision Technology
There are a few choices to be made that will dictate the performance of the PreVision scanning process. We will start by answering the following questions.
- What defect or feature size are we trying to locate using the vision system? This will determine the camera and lens needed.
- What is the total acceptable time that can be used to locate these features? This will determine if the detectable feature size can be achieved.
- How tall, or how much variability in height, do the features/defects have? This will determine how much Optical Depth of Field is needed.
Here is a table showing some of the available options that can be selected based on the answers to the questions above. Note that these 3 performance characteristics are largely mutually exclusive. For example, if you are prioritizing the ability to pick up small feature sizes the total imaging time will be longer, as the FOV will be smaller. Note: There are many more camera and lens options available. The table below represents a relatively small number of these options.
| Vision System Option | Minimum Defect/Feature Detection Size¹(Individual Pixel Size) | Field of View (FOV) | Single-Frame Optical Depth of Field² | Time to Scan 300 x 300mm Area |
| Area Scan Vision Systems | ||||
| Keyence VS-G Area Camera CA-HX500M Camera with CA-LMHR40Lens | 2.58 µm(0.86 µm) | 2.11 x 1.77mmmm | ~0.12mm | ~230 minutes |
| Keyence VS-G Area Camera CA-HX2100M Camera with CA-LMHR13 Lens | 7.5 µm(2.5 µm) | 11.21 x 8.50mm | ~0.45mm | ~15 minutes |
| Keyence VS-G Area Camera CA-HX500 Camera with CA-LMHR05 Lens | 20.7 µm(6.9 µm) | 16.90 x 14.10mm | ~2.15mm | ~7.5 minutes |
| Line Scan Vision Systems | ||||
| Keyence XG-X Line Scan CA-HL08MX Camera with CA-LHL15 Lens | 7.0 µm(2.33 µm) | 19.11mm line Width | ~0.55mm | ~35 minutes |
| Keyence XG-X Line Scan CA-HL04MX Camera with CA-LHL05 Lens | 42 µm(14 µm) | 57.3mm line Width | ~5.12mm | ~3 minutes |
| Digital Microscope | ||||
| Keyence VHX FI Head with 500x Magnification | 0.63 µm(0.21 µm) | 0.61 x 0.46mm | 0.06mm – 0.23mm | Not feasible (>200hrs and 50k images) |
| Keyence VHX FI Head with 50x Magnification | 6.36 µm(2.12 µm) | 6.1 x 4.6mm | 0.08mm – 0.30mm | ~ 170 minutes (~5000 images) |
| Keyence VHX FI Head with 20x Magnification | 15.87 µm(5.29 µm) | 15 x 11mm | 0.43mm – 1.5mm | ~ 30 minutes(~800 images) |
1 – Minimum size is estimated by taking the pixel size and multiplying it by 3. This will give us 9 pixels per area which is a rule of thumb to resolve a feature/defect.
2 – Depth of field is the maximum surface height that is in focus of the camera during any single image capture. This is effectively the max height of the feature/defect that can be detected by the camera, including any height variation across the part’s surface
Practical Considerations and the Need for Testing
In all cases, we suggest application testing prior to the final selection of the vision system. The resolution of the camera, lenses, and lighting all have an effect on the ability to resolve the defects/features that we are searching for. Furthermore, your surface’s unique characteristics will impact the selection as well. How specular (reflective), the color, and the roughness/texture are other factors that impact the performance of the vision system. Testing samples throughout this selection process is the best way to make sure that the images produced will allow for automated detection.
Classification Tools (Software for Feature/Defect Identification)
Each of the main hardware options in the selection section comes with their own defect detection suite of tools.
| Hardware Options | Defect Identification Software Suite |
| Keyence VS-G | The Keyence VS-G series combines advanced AI-driven algorithms with traditional rule-based image processing tools to instantly spot and categorize production flaws. Its AI detection and segmentation tools automatically learn surface characteristics—such as scratches, stains, or burrs—from a few reference images, isolating and classifying complex, variable defects without manual threshold tuning. These are complemented by classic edge, color, pattern-matching, and geometric measurement tools, allowing the system to verify precise physical dimensions alongside flaw classification. |
| Keyence XG-X | The Keyence XG-X series utilizes a customizable, high-speed processing suite designed around flexible flowcharts and programmable algorithms rather than predefined tools. For defect identification and classification, it combines deep learning defect filters—which extract subtle surface anomalies like contamination or dents—with advanced optical processing techniques like LumiTrax multi-directional lighting and 3D height-map inspection. Captured defects are then automatically categorized based on custom user criteria using branching logic, spatial analysis, area thresholds, and statistical feature-grouping tools. |
| Keyence VHX(Paired with MERLIC) | MVTec MERLIC relies on an intuitive, no-code tool suite powered by HALCON algorithms that combines interactive deep learning with traditional machine vision. Its defect identification centres on AI-driven “Detect Anomaly” and “Classify Image” tools, which train on target images without complex coding to automatically flag unseen flaws like scratches or dents and assign them to custom quality categories. These deep learning modules work alongside classic rule-based inspection tools—such as surface texture inspection, edge detection, and pattern matching—to seamlessly sort parts, measure dimensions, and verify defect types in real time. |
Machine Examples
See It In Action – Video Library
conclusion
Relying on an operator to find the inspection locations manually is not scalable in many cases. Adding a vision system to find the inspection areas of interest is an effective way to automate the 3D profilometer’s inspection process. Effectively reducing the cost to run the tool over time, and also the amount of skilled labor that is required to operate it. Selecting the correct vision system is a critical path to automation success.
Reach out to us for consultation so that we can guide the selection process and ensure the highest performance from the combined vision and profilometry system.