Subcategory:
Two camera views can both show a clean circular edge, yet refer to different physical features. A counterbore rim in the upper image and a through-hole exit in the lower image are not automatically the same circle. When we prepare a routine for a Vertical and Horizontal Integrated Instant Vision Measuring Machine, we first establish what each view is looking at. Reliable feature identity is a practical requirement before we combine results or ask an operator to act on them.
Our demonstration shows a HANDING system with vertical and horizontal optical views. The application principle discussed here is identifying the intended feature in each image. The video is not a demonstration of the unrelated photogrammetry system mentioned below.
Multi-camera measurement also appeared in our latest industry review. InLine3D describes a photogrammetric approach that uses images from different angles to determine feature positions. Its technology highlights why the connection between an image feature and a physical point matters. Multiple optical views alone do not establish that a particular measuring machine performs photogrammetric reconstruction; that capability must be confirmed for the actual system.
Start with a feature map, not camera numbers
Camera 1 and Camera 2 are useful equipment labels, but they do not explain the workpiece. We give the drawing features stable names: through-hole A, counterbore A, shoulder B and slot end C, for example. Beside each name we record the intended surface or edge, the view in which it is measurable and the dimensional result it supports. A small annotated image is often clearer than a long instruction.
This mapping remains useful on an Automatic Video Measuring Machine when the routine captures several areas in sequence. The image order should not silently define the feature identity. If a programme is edited and an acquisition moves earlier or later, the report must still identify the same drawing feature. We check that connection explicitly during programme review.
A similar outline may belong to another surface
A stepped bore is a simple example. The upper rim, the bottom of the recess and the exit of the smaller bore can create different circular boundaries. In a second view, one boundary may disappear while another becomes prominent. We would not accept a result merely because a circle tool returned a plausible diameter. We need to know which boundary supplied the points and whether it is the boundary specified on the drawing.
We make the same distinction for a shoulder seen from the side. A silhouette may show the outer envelope while a surface image reveals a groove or chamfer. Those observations can be complementary without being interchangeable. When a measurement requires two features, both feature definitions belong in the programme description, along with the view used to acquire each one.
For a Desktop Instant Vision Measuring Machine, repeated features can create a related problem within one image. A row of similar holes may produce several plausible matches. We therefore include a recognisable orientation reference or other suitable identifying geometry, subject to the available software functions. The nearest-looking hole is not an adequate identity rule when several holes can occupy the search region.
Test the ambiguous positions deliberately
A successful demonstration with one neatly placed sample can hide the weakness. We include placements near the permitted orientation and position limits, along with realistic surface and visibility variations. If the component is symmetrical, we examine whether a rotation or reversal could exchange feature labels without making the image look obviously wrong. The aim is to expose ambiguous associations while the programme is still being qualified.
We also test missing or obscured features. A routine should not silently substitute a neighbouring hole because the expected one cannot be measured. The appropriate response may be a no-result condition or a review request, depending on the validated software and production procedure. We confirm the available behaviour in a trial rather than assuming that a confidence score or a green result guarantees correct identification.
Keep identity and coordinate calculations separate
In a Vision Measuring System, correctly identifying a feature is necessary but does not by itself validate a distance between views. That calculation may also require appropriate calibration, coordinate transformation and a defined dimensional method. We establish those separately. Conversely, a well-calibrated system can still produce an irrelevant number if its routine selects the wrong physical edge.
This distinction makes troubleshooting more productive. When two views disagree, we first compare their annotated images. Are they showing the entrance and exit of a tapered hole? Are they observing different edges of a chamfer? Are they reporting different components of a dimension? We resolve those questions before changing calibration values to force agreement between results that may describe different things.
The benefit of an Instant Vision Measuring Machine is easier to assess once this groundwork is visible. We can evaluate how efficiently the intended features are acquired, what operator intervention remains and which exceptional conditions require review. Acquisition speed has value when the resulting labels and decisions remain connected to the drawing.
Make the trial easy to review
For a buying decision, we recommend retaining a compact evidence set: the marked drawing, view-by-view images, feature labels and the resulting report. Include an explanation of any feature that cannot be observed in the proposed setup. This gives engineering and production staff something concrete to review instead of relying on a demonstration in which a number appears without a visible connection to the part.
We can help assess your application with a suitable Video Measuring Instrument. Send the drawing, photographs of the relevant faces and examples of repeated or easily confused features to handing3d@163.com. At HANDING OPTICS, we use that information to discuss a measurement trial in which each reported feature has an identifiable physical source.
Technical reference: InLine3D technology overview. The reference describes that manufacturer’s photogrammetric method; HANDING model capabilities are assessed separately.
Han Ding offers a one-stop service from needs analysis, product solutions, usage guidance, to after-sales tracking
Address:No. 98, Zhen'an East Road, Chang'an Town, Dongguan City, Guangdong Province, China
Mobile:+86-130 3887 8595 / Aico Ding
Email:handing3d@163.com