PSNR and SSIM Image Quality Metrics Quiz

Answers and explanations appear as you take the quiz.

1. What error quantity underlies the standard PSNR calculation?
   • Mean absolute pixel error
   • The maximum absolute pixel error
   • Mean squared pixel error
   • The fraction of pixels that match exactly
2. What does higher PSNR indicate under the same data-range convention?
   • Higher annotation agreement
   • Lower mean squared reconstruction error
   • More accurate object detection
   • Better color calibration in all cases
3. Which local properties does SSIM compare?
   • Only the global mean pixel value
   • Only independent squared differences at each pixel
   • Only geometric edge coordinates
   • Luminance, contrast and structure
4. Why must PSNR's data range be explicit when comparing implementations?
   • Its peak value depends on the image representation
   • It determines the SSIM window size
   • It removes spatial alignment needs
   • It decides which image is the reference
5. Why can a shifted edge reduce PSNR even when it looks visually similar?
   • Matching resolution makes spatial alignment irrelevant
   • PSNR automatically corrects every spatial offset
   • Pixelwise differences penalize misalignment
   • PSNR compares object identity without considering pixel positions
6. Does a high SSIM guarantee that an image is useful for a specific task?
   • Yes; SSIM validates every downstream decision
   • No; task-specific details may still be lost
   • Yes, if the image is grayscale
   • No; SSIM has no numerical meaning
7. What should be held consistent for a fair PSNR/SSIM comparison?
   • Using the same nominal resolution is enough even if images are misaligned
   • Using the same file format fixes dynamic range differences
   • Averaging scores per image and over pixels always gives the same result
   • Alignment, range, crop, color handling and aggregation
8. What does PSNR report when mean squared error is zero under its idealized formula?
   • An infinite value
   • Zero
   • Exactly one decibel
   • A value equal to the image's peak pixel intensity