4
\$\begingroup\$

I intend to take some raw images using my Canon-450D, and process them offline using some tool. Then I need to get some objective(numerical) metrics for those images like Image Noise, Image Sharpness(blur), Colour Accuracy(or some kind of Colour artifacts) which would help me decide the quality of the processed image.

  1. What are the tools which could help me get these metrics for BMP/JPEG images?It could be 'absolute measurements' or some relative computations which may need a reference image. Either of these options are ok.
    • a. Noise in Image
    • b. Image Sharpness(Blur in the image
    • c. Chroma noise/some metric measuring colour artifacts ?
  2. What are other objective metrics which help in deciding image quality?

P.S.: I have checked Imatest. But there, to get almost any metric, one needs to have the captured images of Standard Test charts (like X-Rite colour chart, ISO Charts, SFRPlus charts, etc.), because it seems to allow the user to select patches in the test chart images as Region of Interest and then computes the metrics for those regions.

It doesn't seem to give metrics for images having any general content/scene in it.

\$\endgroup\$
2

1 Answer 1

1
\$\begingroup\$

Although you're looking for computational metrics rather than human perception ones, I think Joanne C's answer to that question nails both counts. See:

What parameters can I use to evaluate a "Perceptual Image Quality"?, answer #10431.

I'm marking this answer Community Wiki since it's not a direct answer — I encourage everyone to go read Joanne's answer.

(And I'm also not marking this question as a duplicate of that one, since while I think the answer is the same in this sense, there may be pragmatic just give you what you're asking for rather than disagreeing with the assumptions answers that differ from the answers from the human parameters.)

\$\endgroup\$
1
  • \$\begingroup\$ I think the question stands on its own, as the objective quality of an image has a meaning outside of the perceived quality. In particular, the perceived image quality is of little relevance when you're looking for a source image the the most absolute information. As I'm going through old images, I often find two versions of the same image, one with (say) double the resolution, but 90% the file size. A perception test (at least not a very well informed one) may not tell me which one has the highest quality raw information available. \$\endgroup\$
    – Flimzy
    Commented Sep 15, 2015 at 21:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.