3

I want to compare an original video with the same video but transferred through a device. How can I do that? I want to compare them to improve the transferred video quality.

The video is transferred through a device that controls video cameras in operation rooms. It's connected to the monitor through HDMI and so whenever a surgent is doing an endoscopy surgery it'll be shown on the monitor but first transfered from this device.

We want to improve the quality of the video if it's not at it's highest quality. To know that, we should compare it with the original video that is directly connected to the monitor.

I have some parameters in mind like brightness, contrast, frame rate, blurriness,... and many other parameters that can affect the video quality.

How can I measure this?

  • Could you be more specific about "measure everything"? What does "everything entail? – Crowder Mar 25 '15 at 21:52
  • for example measure the frame rate, Color depth, Color Temperature, Resolution, Brightness, Contrast, Saturation, and so on; important parameters in video Comparision – Farah13m Mar 26 '15 at 9:41
  • Got ya. Is there a particular format your original is in? Would you be wanting to compare, say, original film from a film camera with an H.264 digital copy you made from it? I.e., what are the limits of the use case? – Crowder Mar 29 '15 at 2:38
  • No there is no particular format and we can neglect the format it's not very important in my case of study, the most important are parameters like color temperature, white balance, Saturation, Frame Rate, Resolution, Beightness, and other parameters that are enough to use while comparing to get the highest quality, i mean i want sth a device or method that if i compared the two videos with will give me enough informations to decide if the transferred video has the same quality of the original one or not. Thank you so much for answering – Farah13m Mar 29 '15 at 17:18
  • 1
    Since you wrote me an e-mail, I pasted my response (with a few changes) below so others can find it. – slhck Apr 9 '15 at 11:37
2

Like slhck says, you can compare 2 digital video files with metrics like SSIM.

Your question also asked about brightness and contrast, and other things which will depend on the physical monitor that the video is displayed on. If you want to be sure that the right photons come out of the screen, given the pixels it was sent, then you need to calibrate your display. Google that.

You mentioned using HDMI to carry video signals. You can relax here: HDMI is literally lossless (no compression at all, and digital, not analog). 8bit is the standard color depth for video, so you're not losing anything here. It carries at least 8 bits per component (red/green/blue), at resolutions and frame rates high enough for almost any display. (There are different versions of HDMI. Almost all monitors support an HDMI version recent enough to drive them at 60Hz or greater, at their native resolution.)

The question isn't really clear about where in the chain from camera to monitor the 2 different video signals are extracted. Is your "device that controls video cameras" something like a splitter / relay / output duplicator? (not sure what the proper term would be for one-input, several-output-copies device would be, but probably splitter.) If it's just an HDMI splitter, it's almost certainly perfectly lossless.

5

There are quite a few programs that do so-called "full reference" measurement, i.e. comparing an original against a degraded signal:

These are all outputting some value that you can use to objectively compare different systems.

However, keep in mind that the results you get from such a software may not correspond exactly to what humans would rate. These metrics have been trained on natural images and movie sequences and the results may not be valid for medical purposes.

I would suggest you try VQM (it is standardized by the ITU) and image metrics like SSIM. They perform very good, in principle. Then check whether the results are what you would expect when looking at the videos yourself. You could do a small test with a few experts and let them rate a few sequences, then compare the results with the objective values you obtained from the software.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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