12+ Image Quality Metrics that Impact Computer Vision from the Embedded Vision Summit 2022
Have you wondered how Image Quality Impacts Computer Vision?
This content was started for an Embedded Vision Summit talk on May 18, 2022 in Santa Clara, CA.
Quicklink to this page: commonlands.com/summit2022
The Image Quality and Computer Vision Topics
- Camera Exposure and Computer Vision
- Motion Blur
- High Dynamic Range
- Resolution and Sharpness
- Shading and Vignetting
- Noise
- Fisheye Distortion and Wide Angle Lenses
- Fringing and Chromatic Aberration
The Single Pixel Attack contextualizes why Image Quality matters
Have you have heard of the single pixel attack? This is where researchers fooled a simple classification network by changing a pixel.
This adversarial research turned a ship into a car, a horse into a frog, a deer into an airplane!!
Su & Vargas. "One pixel attack for fooling deep neural networks."
While this research is independent from embedded vision systems, it's not a far reach to imagine something similar in the real world. Other researcher have extended this effort to the field and we quote:
“Nonetheless, we show some examples of situations where nearly imperceptible image modifications can result in dramatic perception changes.
Even in applications without malicious people trying to trick your system, the natural world, [your camera hardware, and image processing pipeline] may be adversarial enough.”
Pezzementi, et. al. NREC, “Putting Image Manipulations in Context: Robustness Testing for Safe Perception”
Motion Blur
Please view the motion blur webpage with this content and more.
Dynamic Range and HDR Artifacts
Please view the high dynamic range webpage with this content and more.
Noise
Please view the noise webpage with this content and more.
Color and White Balance
Mahmoud Afifi, Michael S Brown "What Else Can Fool Deep Learning? Addressing Color Constancy Errors on Deep Neural Network Performance "
Tone Mapping
Yeganeh and Wang. Objective Quality Assessment of Tone-Mapped Images
Shading (Vignetting)
Resolution (Sharpness, Angular)
Please view the resolution webpage for this content.
Image Compression
Allen. "Image Quality Evaluation in Lossy Compressed Images."
Zheng, et. al. “Improving the Robustness of Deep Neural Networks via Stability Training”
Distortion (and Angular Resolution)
Please view the wide angle and fisheye distortion webpage for this content.
Texture Blur/Resolution
Please view the noise webpage with this content and more.
Stray Light
IEEE P2020 White Paper
Color Fringing
Kang "Automatic Removal of Chromatic Aberration from a Single Image"
More on Blooming from Imatest
Blemishes
Urcar "Let’s Get Dirty: GAN Based Data Augmentation for
Camera Lens Soiling Detection in Autonomous Driving"
More on Blemishes from Imatest
Looking for recommendations, trying to figure out where to start, Or Want to Discuss Customizations?
Contact our US-based Optical Engineers.
Working on a camera hardware project? What's your application?
Surveillance?
Our low F#, high resolution lenses are suitable for 180° dome cameras, low light, and active IR illuminated scenes.
Aerial Robotics?
Our light-weight, miniature lenses are ideal for collision avoidance, long-distance viewing, and end-customer viewing.
Does it Require an IP65+ Lens?
We offer IP65+ variants of many lenses. These are suitable for applications exposed to the environment, without a window.
Surveillance?
Our low F#, high resolution lenses are suitable for 180° dome cameras, low light, and active IR illuminated scenes.
Consumer / AR+VR?
Our high resolution fisheye lenses will let you design 360° cameras and and stereographic lenses are equivalent to "GoPro" lenses.
Video Conferencing?
Our wide-angle, low distortion, high resolution M12 lenses result in optimal object deformation without post-processing.
Video Conferencing?
Our wide-angle, low distortion, high resolution M12 lenses result in optimal object deformation without post-processing.