Sometimes less is more

Your field of view should not change but the frame rate should be increased? You need a shorter exposure? Will a lower resolution with the same field of view do for you?

To reduce the image resolution and still having the same field of view, you have different ways depending on the sensor model: subsampling, binning, or scaler. But how do they differ?

Background

Subsampling

Subsampling skips multiple sensor pixels when reading out the image data. This reduces the amount of data to be transferred and enables higher camera frame rates. The captured image has a lower resolution but still the same field of view compared to the full-resolution image.

Color subsampling – as performed by most color sensors – skips pixels while maintaining colors. For some monochrome sensors, the camera also performs color subsampling, resulting in slight artifacts.

Monochrome sensors and some color sensors ignore the Bayer pattern and the color information gets lost (mono subsampling).

Subsampling can also be done in the pixel preprocessing on the camera. This process reduces the data volume per image, but does not increase the frame rate.

Binning

Binning is a function that averages or adds multiple sensor pixels to obtain a sin-gle value:

  • If the pixel values are added, the image brightness increases.
  • If the pixel values are averaged, the image noise is reduced.

This also reduces the amount of data to be transferred and enables higher cam-era frame rates. The captured image has a lower resolution but still the same field of view compared to the full-resolution image.

Color binning – as performed by most color sensors – combines only pixels of the same color. For some monochrome sensors, the camera also performs color binning, resulting in slight artifacts.

Most monochrome sensors and some color sensors combine neighboring Bayer pattern pixels; in this case, the color information gets lost (mono binning).

Setting the image resolution in uEye Cockpit

Open the camera properties in uEye Cockpit via “uEye > Properties”. Depending on the opened camera model, you can configure subsampling, binning, or scaler in the “Size” tab.

Tip: Image profiles

If the field of view or the size of the image detail is allowed to be changed, you can use alternatively an image profile.

In the “Size” tab, you can set a predefined resolution via a profile. Depending on the profile and sensor model, the camera driver sets automatically the requested size via AOI, binning, subsampling, or scaler to receive the best image quality.

Using image profiles

Application areas

All three methods have in common that the field of view may not be changed but the image resolution may be reduced.

The advantage of subsampling is that the frame rate can be increased without changes in image brightness or contrast. The scaler allows the same as subsampling but with finer intermediate steps. However, the scaler is only supported by certain sensors.

The advantage of binning is that image brightness can be increased. This advantageous in applications in which a flash cannot be used for example for light-sensitive samples or reflective surfaces. Also you can use a shorter exposure time which is important for moving objects.

Summary

Subsampling

Binning

Scaler

Field of view

unchanged

unchanged

unchanged

Image resolution

reduced

reduced

reduced

Frame rate

may increase

may increase

may increase

Image contrast

unchanged

may increase

unchanged

Image brightness

unchanged

may be brighter

unchanged

Image noise

unchanged

may be better

unchanged

Advantage


  • Data reduction

  • Can be set independently in vertical or horizontal direction


  • Data reduction

  • Can be set independently in vertical or horizontal direction

  • In certain circumstances shorter exposure times are possible

  • Binning is superior to subsampling in terms of image quality


  • Data reduction

  • Reduces the image almost continuously

Disadvantage


  • Pixel details get lost

  • Fixed step widths


  • Fixed step widths


  • Only available for few sensor models