Computer Vision

• Introduction to Computer Vision
  • What is Computer Vision
    • Computer Vision vs Computational Photography
  • Why Study Computer Vision
  • Why Computer Vision is Hard
    • Adelson’s Checker Shadow Illusion
    • Perception is Active Construction
  • Course Structure
    • Topic Outline
• Images as Functions
  • Images as 2D Functions
    • Bounded Domain and Range
    • Color Images
  • Digital Images
    • MATLAB/Octave Basics
  • Image Arithmetic
    • Addition and Averaging
    • Scalar Multiplication
    • Alpha Blending
    • Image Subtraction
  • Noise Models
    • Common Types
    • Generating Gaussian Noise
    • Effect of Sigma
  • Displaying Images
• Filtering: Cross-Correlation, Convolution, and Sharpening
  • Linearity of Operators
    • Impulse Function and Response
  • Correlation vs Convolution
    • Properties of Convolution
  • Computational Complexity and Separability
  • Boundary Issues
  • Sharpening Filters
  • Median Filter
• Filtering: Gaussian Smoothing, Noise Removal, and Edge Effects
  • Noise Removal by Averaging
    • Weighted Moving Average
    • 2D Moving Average
  • Cross-Correlation
  • Gaussian Filter
    • Parameters
    • MATLAB/Octave
    • Two Sigmas: Noise vs Smoothing
    • Gaussian Filter Properties
• Template Matching
  • Normalized Cross-Correlation
    • 1D Correlation for Matching
  • 2D Template Matching
    • MATLAB/Octave Implementation
  • Limitations of Template Matching
    • Non-Identical Matching
• Edge Detection: Gradients and Derivatives
  • Edges in Images
    • Types of Edge Boundaries
  • Edge Detection via Derivatives
  • Image Gradient
    • Finite Differences
    • Sobel Operator
    • Gradient Direction
  • Dealing with Noise
    • Second Derivative and Zero Crossings
• Edge Detection: Canny Operator and 2D Methods
  • Derivative of Gaussian in 2D
    • Effect of Sigma
  • Canny Edge Detector
    • Non-Maximum Suppression
    • Hysteresis Thresholding
    • Canny and Noise
  • Laplacian of Gaussian (LoG)
    • MATLAB Edge Detection Summary
• Hough Transform: Line Detection
  • Parametric Models
  • Challenges of Line Fitting
  • Voting and the Hough Transform
    • Hough Space (m-b Parameterization)
    • Polar Representation
  • Hough Transform Algorithm
    • Complexity
  • Examples
    • MATLAB Demo Summary
    • Line Segments vs. Infinite Lines
  • Noise and Robustness
  • Extensions
    • Gradient-Based Voting
    • Other Extensions
• Hough Transform: Circle Detection
  • Circles with Known Radius
  • Circles with Unknown Radius
    • Using Gradient to Reduce Voting
  • Algorithm
  • Practical Tips
  • Pros and Cons
• Generalized Hough Transform
  • Non-Analytic Models
  • R-Table Construction (Training)
  • Detection (Known Orientation and Scale)
    • Handling Unknown Orientation
    • Handling Unknown Scale
  • Visual Codeword Approach
    • Training
    • Detection
• NumPy Reference
  • Array Operations
  • Log-Sum-Exp
  • Resources
• OpenCV
• Images
  • Reading and Displaying Images
• Terms