Why a Telecentric Lens Produces a Double Image on Polished Metal

What the double image actually is

On a polished metal surface the sensor receives two overlapping images of the same edge. One is formed by light that scatters from the edge itself. The other is formed by light that leaves the surface specularly, travels to a nearby mechanical part, bounces back, and re-enters the lens. Because both paths reach the same pixels, the edge appears twice with a small offset. The offset is a function of the angle between the surface and the reflecting part, not of the lens magnification, which is why it does not scale when you change lenses.

How to tell it apart from a lens defect

Three checks separate a reflection artifact from an actual optical fault. First, rotate the workpiece in its own plane while keeping it at the same height: a reflection artifact moves with the part, a lens defect stays fixed relative to the field of view. Second, change the working distance by a few millimetres: a reflection artifact barely shifts because the bounce geometry is unchanged, while a defect inside the lens stack stays in the same pixel position. Third, block the suspect surface with a matte card without touching the part: if one of the two edges disappears, the cause is a reflection.

Why telecentric designs are more exposed to it

A telecentric lens accepts a narrow cone of light from each field point. That is exactly what makes magnification stable across height, but it also means the lens only accepts rays that leave the surface almost parallel to the optical axis. Specular reflections from polished metal often satisfy that condition over the whole aperture, so the reflected image arrives with almost full intensity and is imaged as sharply as the real edge. A standard lens accepts a wider cone, so much of the reflected light falls outside the aperture and the effect looks weaker, which is often misread as the telecentric lens being worse.

Optical fix: move the specular lobe out of the aperture

The practical rule is to keep the specular lobe outside the acceptance cone. With a POMEAS telecentric lens covering fields from 0.4 mm to 265 mm and working distances from 9.5 mm to 468 mm, the choice of model determines how much room you have. A longer working distance gives more space to tilt the illumination without moving the lens, which is why the 398 mm and 416 mm models are often used on reflective metal even when the field of view would allow a shorter model.

Geometry fix: remove the second reflector

Most double images are produced by a second surface, not by the workpiece. Typical culprits are the inner wall of a fixture bore, the polished chamfer of a locating pin, and the anodised surface of the fixture plate itself. Once the second reflector is removed or roughened, the ghost disappears without any change to the lens. This is why the first diagnostic step is to map every surface that can see the field of view, not to change the lens.

What the measured data allows you to specify

When you write the inspection requirement, quote the resolution and distortion of the specific model rather than the series. Across the POMEAS telecentric range, object-space resolution spans 1.1 to 398 micrometres and TV distortion spans 0.002% to 0.06%. A model at the coarse end of the resolution range will not resolve a ghost edge separated by less than its resolution, so the ghost may appear as a grey smear rather than a second edge. Selecting a model whose resolution is well below the expected ghost separation makes the problem visible and therefore fixable.

Frequently asked questions

Is a double image on metal caused by the telecentric lens itself?

No. In most cases a second surface in the field of view reflects light back into the lens. Rotating the workpiece distinguishes a reflection artifact from a lens defect.

Why does the double image not change when I change the lens?

The offset is set by the angle between the polished surface and the reflecting part, not by magnification. Changing lenses changes scale but not that angle.

Does a longer working distance reduce the ghost?

It gives more physical room to move the illumination outside the specular lobe, which usually removes the ghost. POMEAS telecentric models span 9.5 mm to 468 mm working distance.

Can I remove the ghost with software?

Only if the two edges are separated by more than the model resolution. If they overlap below the resolution limit, no filter can recover the true edge.

What should I measure before choosing a model?

The expected ghost separation and the part height variation. Resolution and telecentricity then determine whether the ghost is resolvable and whether height changes matter.

Which POMEAS models are typically used on polished metal?

Long working distance models such as the 398 mm and 416 mm versions, because they leave room to angle the light away from the specular direction.