Readings: Image Formation

CSC 262 - Computer Vision - Weinman

Summary:

We briefly direct your attention to the most important parts of the assigned readings on image formation.

Todays readings cover a wide variety of ideas you'll want a working familiarity with in order to understand the (literal!) ins and outs of the images you work with. To be most efficient, use this guide as you read to help you skim what can be skimmed and mark the high points you'll want as takeaways.

Trucco & Verri (Chapter 2: Digital Snapshots)

2.1 Introduction

This section is concise and to the point. Internalize the two consequences of "image ≈ matrix" to motive why you'll read on about image formation.

2.2 Intensity Images

Focus on the construction, connecting the words to the figures. While most mathematics here can be skimmed (focusing on the prose explanations/intuitions), there are exceptions:

• don't miss the thin lens equation (2.2)
• we'll examine the implications of Eq. (2.13) in class - it's very cool! do not worry about any of the mathematical derivations on pp. 23-25, but do try to understand where the terms of Eq (2.13) show up in Figures 2.5 and 2.7
• you do need to know the fundamental perspective equations (2.14), which we'll derive (for intuition) in class

2.3 Acquiring Digital Images

Figure 2.9 gives the overview, but skip pp. 30-31. Section 2.3.3 is essential reading, however, because you'll be implementing the EST_NOISE algorithm and examining the implications for noise in lab. (AUTO_COVARIANCE is less important and may be skimmed.

2.4 Camera Parameters

We'll return to many ideas from this reading in later weeks, but here are the big ideas to focus on

• Figure 2.13 gives you the intuition for Eq. (2.19)
• We'll unpack Eq. (2.20) in class

You can stop there with Eq. (2.20), but before you go on to Szeliski's reading, turn the page and read the summary of Intrinsic Parameters atop p. 38.

Szeliski (Chapter 2: Image Formation)

Color filter arrays

Make a mental note about Figure 2.31, because it may come up in the lab!

Color balance

Skip this section

Gamma

We will explore the gamma correction in the next lab, but for now it helps you consider the arbitrary relation between sensor voltage and recorded brightness.

Do not go on to Other color spaces.