Part IV: Human Vision
The book is still taking shape, and your feedback is an important part of the process. Suggestions of all kinds are welcome—whether it’s fixing small errors, raising bigger questions, or offering new perspectives. I’ll do my best to respond, but please keep in mind that the text will continue to change significantly over the next two years.
You can share comments through GitHub Issues.
Feel free to open a new issue or join an existing discussion. To make feedback easier to address, please point to the section you have in mind—by section number or a short snippet of text. Adding a label characterizing your issue would also be helpful.
Last updated: November 6, 2025
The material about human vision in this volume is currently under development.
For now, please refer to the chapters in Foundations of Vision.
The new material in this volume will emphasize the human vision engineering standards, reducing the emphasis on biology that is part of FOV.
Human Vision Topics
- Seeing
- Human spatial encoding
- Human wavelength encoding
- Image quality metrics
When designing a digital camera, how good does it need to be? The answer depends on the intended use case. If the images are for a person to see, the camera’s design must be guided by the capabilities and limits of the human visual system. This principle has profound consequences for engineering.
Vision scientists have learned a great deal about the initial stages of vision. We have excellent measurements of how the eye’s optics transform light into an image on the retina, and we understand how the retinal photoreceptors encode that image into neural signals. This knowledge is a practical guide for designing any system that creates or displays images for people.
Consider spatial resolution. It would be a disappointment if a camera failed to capture details as fine as those the human eye can resolve. We expect our photos to be at least as sharp as what we see. However, unlike with color, people are often pleased if a camera captures a scene at a finer resolution than the eye can perceive. We can always zoom in on the digital image to see details we might have missed. This extra information does not interfere with the viewing experience; it just requires more storage and processing power.
Next, consider color. A camera designed for human viewing should capture the same portion of the electromagnetic spectrum that we see: visible light. Capturing less of the spectrum would lead to poor color reproduction. Capturing more—like infrared or ultraviolet—would be wasteful, consuming resources to record information we cannot see. In this sense, human wavelength sensitivity sets a clear target for camera design.
These two aspects of human vision provide different quantitative engineering constraints:
- Spatial Resolution: Human vision sets a minimum bar. Capturing more detail can be beneficial.
- Wavelength Sensitivity: Human vision defines a target range. Capturing information outside this range is generally wasteful.
After discussing the general principles of seeing, we will delve into the quantitative properties of human vision that guide engineering decisions for spatial and color reproduction. We will then describe an array of image quality metrics that are based on these properties and are commonly used in commercial applications.