Reliable way of measuring digital camera image noise?

[DanielJ]

Is there a basic and reliable method for measuring noise levels in a digital capture? Sure would be nice to quantify any image quality deterioration.

[McQ]

The most inclusive way to quantify noise is a "standard deviation vs. tone" plot, for each color channel and for luminance. It does not account for noise frequency (only magnitude), but it'll do just fine for most purposes. The only problem is that this plot does not account for excessive in-camera noise reduction, so a camera manufacturer could effectively "cheat" and do exceedingly well on this plot--even though the photos themselves would contain little detail. Unfortunately the best way is therefore subjective IMO, where one looks at the photos themselves and draws their own conclusion.

That being said, if you are interested with relative noise *for the same camera*, or for cameras where you are confident in-camera noise is minimal, then a standard deviation versus tone plot would be just fine. You can create one manually in photoshop by selecting equal sized patches of various shades of gray (using the Histograms palette in the "expanded view" mode).

The histogram applies to the selected portion of region 2 above. Notice how the standard deviation is listed below (as "Std Dev"). For more on this topic, also take a look at the tutorial on digital camera image noise: examples and characteristics. Here's a figure from that page illustrating the concept of standard deviation as it applies to each noise patch:

Alternatively, you could develop something automated using matlab or a photoshop action if you're comfortable with that...