Gary Bishop

F06 Images Graphics Vision Poop

Prerequisites

  • Calculus of single and multiple variables
  • Linear algebra: points & vectors, dot and cross products, orthogonality, linear transformations by matrix multiplication
  • COMP 410: Data Structures

Syllabus

  • Images and objects (1.5 weeks)
    • Forms of discrete representation
      • Sampled
      • Via composition of functions on parameters
    • Noise
    • Blurring and distortion
    • Color and shading
  • Sensing (2.0 weeks)
    • Cameras
      • Light cameras: analog to film, analog electronic, digital, digitizers, cine, Pinhole, lens
      • X-ray and ultrasound imagers
      • Retinal sensors, preprocessing through LGN to V1
        • Receptive fields, spatial scales
        • Spatial and temporal derivatives
    • Resolution in sensing
    • Spatial distortion
    • Sampling and integration, aliasing
    • Form sensing in human vision, structure of the human visual system
  • Display (2.25 weeks)
    • Devices and their operation: CRT & video, intensity & color, color spaces
    • Display as composition of intensity transformation, display device, & perception
    • Intensity & color perception
    • Display linearization: gamma correction, just noticeable differences
    • Contrast & brightness control: intensity windowing, histogram equalization, lookup tables
    • Drawing lines and text
    • Object rendering
      • Normals & shading
      • Reflectivity & illumination: ambient, diffuse, specular reflections
  • Noise, unwanted details, & scale choice in both images and objects (4.0 weeks)
    • Scale: level of detail, aperture, interrelation distances
      • Apertures
        • Global
          • Orthogonal function compositions: Fourier, spherical harmonics
          • Filtering
          • Speedy Fourier analysis: the DFT and the FFT
          • Singular Value Decomposition & Principal Component Analysis
        • Local
          • Gaussian and its derivatives
          • Wavelets
          • Splines
  • Spatial/time sampling interpolation from discrete to continuous (2.0 weeks)
    • Smoothing
      • Weakening small levels of detail
      • Coarser sampling
    • Sampling rate choice and anti-aliasing
    • Multiscale sampling
    • Interpolation, as convolution
  • Selected 3D matters (2.25 weeks)
    • Similarity transformations in 3D
      • By 3D matrices plus vector addition, incl. rotation through Euler angles
      • Rotation via quaternions
      • Orthogonal projection, change of coordinate systems
    • Homogeneous coordinates & transformation matrices
    • 2D images from 3D
      • Perspective & perspective projection
      • X-ray (accumulated projection)

Texts

Computer Graphics: Principles and Practice in C by James D. Foley, Andries van Dam, Steven K. Feiner and John F. Hughes.

F06 Other Readings

Recommended Computing environment

MATLAB. Recommended reference: Mastering MATLAB 7 by Duane C. Hanselman and Bruce L. Littlefield.

If you are bold, brave, and open source oriented, you can do everything you need to do in SciPy.

Honor Code

In all course activities you must give credit where credit is due.

Course Work and Grading.

  • Homework (biweekly)
    • On time is the beginning of class on the day due.
    • Slightly late is before grading for that assignment is complete
    • Very late is after that
  • Midterm exam
  • Final exam

`H_o` = [ on time homework grades ]
`H_s` = [ slightly late homework grades ]
`H_v` = [ very late homework grades ]
`N` = number of homeworks
`M` = midterm grade
`F` = final exam grade

With all grades normalized to 0 – 100%

` Grade = 4/9 (sum H_o + 9/10 sum H_s + 6/10 sum H_v)/N + 5/9 (max(M,F) + F)/2 + Diddly `