MECH5720 Chap.3 Classical Imaging Tradeoffs
Classical Imaging Tradeoffs
A conventional camera cannot optimise exposure, depth of field, diffraction, motion blur, field of view and sampling independently. Aperture diameter and focal length determine the f-number; exposure time trades photon count against motion; sensor pitch sets a sampling limit that may be lower or higher than optical resolution.
This chapter uses controlled comparisons to identify the active bottleneck rather than treating one sharpness score as a complete image-quality measure. Exposure decisions begin with the photons required by the task. Extending shutter duration raises expected count for a static scene, often improving the relative influence of shot and read noise.
The same choice integrates object or camera motion over a longer interval, turning displacement into blur. Analogue gain can brighten the output but cannot recover photons that were never collected, and excessive exposure can saturate highlights whose true value is then lost. Aperture participates in several competing effects.
A wider pupil admits more energy and supports a shorter exposure, while a narrower pupil increases the range that appears acceptably focused and may reduce some aberrations. Very small openings strengthen diffraction, which attenuates fine spatial detail even when the depth range looks improved. Comparisons must keep focal length and exposure compensation visible because changing several controls together obscures the cause.
Sampling creates another ceiling. The sensor only records discrete positions, so optical detail above one half cycle per pixel folds into aliases unless suppressed before sampling. Magnification can spread a target over more pixels, but changes field of view and may require a longer working distance.
Smaller pixels offer denser sampling on a fixed sensor size while receiving fewer photons per pixel at the same irradiance and exposure. Measurement should follow the claimed benefit. A Siemens star or bar target provides known frequencies, a slanted edge estimates transfer across frequency, and repeated natural scenes test whether a setting is robust outside one calibration pattern.
Variance of a Laplacian can rank a controlled focus sweep, yet the score also rewards noise and ringing. It should be paired with visual inspection and a target-based measure. Diagnose bottlenecks by changing one factor. A shorter exposure that repairs an edge on a moving target supports motion blur; a static target captured under the same settings checks defocus.
A rotated or translated fine pattern can reveal phase-sensitive aliasing. Captures at several depths separate central-plane sharpness from useful depth of field. Each conclusion should name the geometry, exposure, aperture, gain and measurement region so the tradeoff can be reproduced rather than described as a universal best setting.
What this chapter covers
- 01
F-number and throughput
- 02
Depth of field
- 03
Diffraction and aberration
- 04
Motion blur
- 05
Nyquist sampling
- 06
Resolution-target measurement
Worked example · free
Identify aliasing before interpolation
- +1Compute the sampled frequency as 22/40 = 0.55 cycles per pixel.
- +1Compare with the Nyquist limit of 0.50 cycles per pixel.
- +1Conclude that the pattern aliases to a misleading lower frequency.
- +1Increase optical magnification or sampling density, or prefilter before sampling.
Key terms
- F-number
- Focal length divided by effective aperture diameter.
- Depth of field
- The range of object distances rendered acceptably sharp.
- Diffraction
- Wave spreading at an aperture that limits fine-detail transfer.
- Aliasing
- Frequency folding caused by sampling above the Nyquist limit.
- Resolution target
- A known pattern used to measure contrast and distinguishable detail.
Classical Imaging Tradeoffs FAQ
Does a smaller aperture always sharpen the image?
Stopping down can reduce aberration and increase depth of field, but it also reduces light and eventually increases diffraction blur. The best setting depends on scene depth, exposure constraints, optical quality and the spatial frequencies needed by the task.
Why can one sharpness score be misleading?
Gradient or Laplacian statistics can increase with real edges, noise or ringing. Pair them with a known target, contrast-versus-frequency measurement and visual artifact inspection before claiming improved resolution.
How should two camera settings be compared?
Change one factor at a time where possible, preserve geometry and target, record exposure and gain, and repeat the capture. A matched control reveals which tradeoff caused the measured difference.
Exam move
For each control, write both the benefit and the cost: aperture, exposure, focal length, pixel pitch and gain. Solve short sampling calculations, then explain whether the remedy belongs before or after capture. Use resolution targets to connect a qualitative artifact to a measurable frequency range. Create comparison cards for motion blur, defocus, diffraction and aliasing.
On each card, write the predicted visual symptom, one variable to change and a control that excludes the closest alternative. Recalculate frequencies after changing magnification or pixel count rather than relying on intuition. When ranking focus settings, state why the metric might reward noise. A complete practice answer should identify the active bottleneck and one new cost introduced by the proposed repair.
Repeat one target capture after a subpixel translation and use the change to test for aliasing.