When working on line-scan projects, one of the most challenging parameters to decide is…Horizontal frequency.

The customer immediately said:

“Our production line runs at 22 m/s. For this project, what line-scan camera with a high frame rate should we choose?”

Actually, there’s no need to start by flipping through the spec sheet.

First, let’s clarify the three parameters:

Width, precision, and speed.

You can basically figure out the camera specs.

Let’s start with this case.

Project Requirements:

• Detection width:1600 mm

• Detection accuracy:1 mm

• Movement speed:22000 mm/s, which is 22 m/s

Step 1: First, calculate how many pixels are needed.

Formula:

Number of pixels = Width ÷ Resolution

So:

1600 ÷ 1 = 1600 pixel

In theory, 1600 pixels is sufficient.

But the cameras all come with standard specs—just step up one level:

Simply select 2K, which is 2048 pixels.

Step 2: How fine can a 2K camera actually see?

Let’s work it backward again:

Actual precision = Width ÷ Number of pixels

That is to say:

1600 ÷ 2048 = 0.78125 mm/pixel

Approximately:

0.78 mm/pixel

Simply put:

One pixel corresponds to approximately 0.78 mm in the real world.

Step 3: Here’s the key—how do you calculate the horizontal scan rate?

Simply put, the horizontal scan rate is:

How many lines does the camera scan per second?

The formula is also very simple:

Line frequency = Motion speed ÷ Motion direction accuracy

The current speed is:

22000 mm/s

If sampling is also performed at 0.78 mm per line in the longitudinal direction:

22000 ÷ 0.78125 ≈ 28160 line/s

That is to say:

Approximately 28.2 kHz

So, in theory, this project should at least select:

2K-resolution, line-array camera with a line rate exceeding 28.2 kHz.

However, in practice, the actual selection should not be constrained by theoretical values.

The theoretical calculation yields 28.2 kHz.

So, would it be fine to just choose 29 kHz?

Generally not recommended.

Because the real scene also includes:

Speed fluctuations,Encoder error, exposure time, trigger delay, and post-processing speedup.

Therefore, in real-world projects, it’s advisable to allow for some margin.

For example, in this case, the following could be considered:

30 kHz, 35 kHz, or even higher scan rates.

There’s another issue that’s particularly easy to get wrong.

The customer said:

“I want to detect 1 mm defects.”

Many people’s first reaction is:

Then let’s calculate it as 1 mm/pixel.

Actually, not necessarily.

Because a 1 mm defect, if it occupies only one pixel in the image, typically results in poor detection stability.

If you want a 1 mm defect to cover 2 pixels:

1 ÷ 2 = 0.5 mm/pixel

That 1600 mm width requires:

1600 ÷ 0.5 = 3200 pixel

At this point, 2K is no longer sufficient; you basically need to consider:

4K-line array camera.

So at the very beginning of the project, you must make sure to ask clearly:

What does the customer mean by “1 mm”—is it 1 mm per pixel, or do they want to reliably detect defects that are 1 mm in size?

These two sentences may seem similar, but the camera ultimately chosen could be completely different.

Finally, remember these three formulas.

Pixel = Width ÷ Resolution

Actual precision = Width ÷ Pixel

Line frequency = Speed ÷ Resolution

Returning to this case:

1600 mm width + 1 mm accuracy + 22 m/s speed

The preliminary results are:

2K-line array camera

Actual accuracy is approximately 0.78 mm/pixel.

Theoretical operating frequency is approximately 28.2 kHz.

Then, a final confirmation is made based on the lens, light source, exposure, encoder, and interface bandwidth.

Selecting a line-scan camera isn’t actually that complicated.

Calculate carefully first, then choose the model.

Especially for high-speed projects,Never choose the line frequency based on guesswork.