How Image Formation Works in Computer Vision: A Beginner’s Guide

📷 Image Formation in Computer Vision (Explained Simply)

Computer vision is about teaching machines to “see.” But before a computer can understand images, it must first learn how images are formed.

This guide explains the entire process—from light to pixels—in a simple, structured way.

---

📚 Table of Contents

• What is Image Formation?
• Step 1: Light and Scene
• Step 2: Camera Lens
• Step 3: Image Sensor
• Step 4: 3D to 2D Projection
• Math Behind Image Formation
• Key Concepts
• Putting It All Together
• Key Takeaways

---

🧠 What is Image Formation?

Image formation is the process where light from a scene is captured and converted into a digital image.

Human eye → Brain processes light Camera → Sensor processes light into pixels

Both systems work similarly: they convert light into interpretable information.

---

🌟 Step 1: Light and the Scene

Everything starts with light.

• Light hits objects
• Objects reflect light
• Reflected light enters camera

The intensity and color of light determine what the image looks like.

If we represent light intensity mathematically:

\[ I(x, y) = \text{light intensity at pixel (x, y)} \]

This means every pixel stores brightness information.

---

🔍 Step 2: Camera Lens

The lens focuses light onto the sensor.

Without a lens → blurry image With proper lens → sharp image

Refraction (Light Bending)

Light bends when passing through the lens.

This bending helps all rays meet at a point called the focal point.

---

📡 Step 3: Image Sensor

The sensor is made of millions of pixels.

Each pixel measures light intensity.

Pixel Function:

\[ Pixel = f(\text{incoming light intensity}) \]

So the image becomes a grid of numbers (matrix).

Example:

[[12, 45, 78], [34, 90, 120], [10, 60, 200]]

This matrix is what a computer actually sees.

---

📉 Step 4: 3D → 2D Projection

A real-world scene is 3D, but images are 2D.

This conversion is called projection.

Mathematically:

\[ (x, y, z) \rightarrow (x', y') \]

Simple Explanation:

A shadow of a ball is 2D, but the ball is 3D.

Same idea applies to cameras.

---

📐 Math Behind Image Formation

1. Pinhole Camera Model

\[ x' = f \cdot \frac{x}{z}, \quad y' = f \cdot \frac{y}{z} \]

Easy Explanation:

• \(x, y, z\) = real-world coordinates
• \(f\) = focal length
• \(x', y'\) = image coordinates

👉 Objects farther away (large z) appear smaller.

---

2. Light Intensity Model

\[ I = L \cdot R \]

• L = light source
• R = reflection from object

👉 Brightness depends on both light and surface.

---

📊 Key Concepts

📏 Focal Length

Controls zoom level of the camera.

👁️ Field of View

How much of the scene is visible.

💡 Aperture

Controls light entering the camera.

🌗 Depth of Field

Range of sharp focus in image.

Shallow depth → blurred background Deep depth → everything sharp

---

⚙️ Putting It All Together

1. Light reflects from objects
2. Lens focuses light
3. Sensor captures light as pixels
4. 3D world becomes 2D image

The final output is a matrix of numbers that represents an image.

---

💡 Key Takeaways

• Images are made of light information
• Cameras convert light into digital pixels
• Mathematics helps map 3D → 2D
• Every image is just a matrix of numbers

---

🎯 Final Insight

Image formation is the foundation of computer vision. Without it, AI systems would not be able to interpret the world visually.

Understanding this process helps in areas like:

• Autonomous driving 🚗
• Facial recognition 😊
• Medical imaging 🏥
• Robotics 🤖